Initial commit

mplug_docowl/model/builder.py

+#    Copyright 2023 Haotian Liu
+#
+#    Licensed under the Apache License, Version 2.0 (the "License");
+#    you may not use this file except in compliance with the License.
+#    You may obtain a copy of the License at
+#
+#        http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS,
+#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#    See the License for the specific language governing permissions and
+#    limitations under the License.
+
+
+import os
+import warnings
+import shutil
+
+from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig, BitsAndBytesConfig
+from transformers.models.clip.image_processing_clip import CLIPImageProcessor
+import torch
+from mplug_docowl.model import *
+from icecream import ic
+def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, load_4bit=False, device_map="auto", device="cuda"):
+    kwargs = {"device_map": device_map}
+
+    if device != "cuda":
+        kwargs['device_map'] = {"": device}
+
+    if load_8bit:
+        kwargs['load_in_8bit'] = True
+    elif load_4bit:
+        kwargs['load_in_4bit'] = True
+        kwargs['quantization_config'] = BitsAndBytesConfig(
+            load_in_4bit=True,
+            bnb_4bit_compute_dtype=torch.float16,
+            bnb_4bit_use_double_quant=True,
+            bnb_4bit_quant_type='nf4'
+        )
+    else:
+        kwargs['torch_dtype'] = torch.float16
+    if 'paperowl' or 'docowl' in model_name.lower():
+        if model_base is not None:
+            # this may be mm projector only
+            print('Loading mPLUG-DocOwl from base model...')
+            tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False)
+            cfg_pretrained = AutoConfig.from_pretrained(model_path)
+            model = MPLUGDocOwlLlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=cfg_pretrained, **kwargs)
+        else:
+            tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False)
+            model = MPLUGDocOwlLlamaForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs)
+    else:
+        # Load language model
+        if model_base is not None:
+            # PEFT model
+            from peft import PeftModel
+            tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False)
+            model = AutoModelForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, **kwargs)
+            print(f"Loading LoRA weights from {model_path}")
+            model = PeftModel.from_pretrained(model, model_path)
+            print(f"Merging weights")
+            model = model.merge_and_unload()
+            print('Convert to FP16...')
+            model.to(torch.float16)
+        else:
+            use_fast = False
+            tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False)
+            model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs)
+
+    # vision_tower = model.get_model().vision_model
+    # vision_tower.to(device=device, dtype=torch.float16)
+    # image_processor = CLIPImageProcessor.from_pretrained(model_path)
+    image_processor = None
+
+    if hasattr(model.config, "max_sequence_length"):
+        context_len = model.config.max_sequence_length
+    else:
+        context_len = 2048
+
+    return tokenizer, model, image_processor, context_len
\ No newline at end of file

mplug_docowl/model/configuration_mplug_docowl.py

+# Copyright (c) Alibaba.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+import copy
+import os
+from typing import Union
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.models.auto.modeling_auto import MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
+from transformers.utils import logging
+from transformers.models.auto import CONFIG_MAPPING
+
+
+class LlamaConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the LLaMA-7B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 32000):
+            Vocabulary size of the LLaMA model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`LlamaModel`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 11008):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer decoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        num_key_value_heads (`int`, *optional*):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 2048):
+            The maximum sequence length that this model might ever be used with. Llama 1 supports up to 2048 tokens,
+            Llama 2 up to 4096, CodeLlama up to 16384.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        pad_token_id (`int`, *optional*):
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            End of stream token id.
+        pretraining_tp (`int`, *optional*, defaults to 1):
+            Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
+            document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
+            necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
+            issue](https://github.com/pytorch/pytorch/issues/76232).
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
+            strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
+            `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
+            `max_position_embeddings` to the expected new maximum. See the following thread for more information on how
+            these scaling strategies behave:
+            https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
+            experimental feature, subject to breaking API changes in future versions.
+        attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
+
+
+    ```python
+    >>> from transformers import LlamaModel, LlamaConfig
+
+    >>> # Initializing a LLaMA llama-7b style configuration
+    >>> configuration = LlamaConfig()
+
+    >>> # Initializing a model from the llama-7b style configuration
+    >>> model = LlamaModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+    model_type = "llama"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=32000,
+        hidden_size=4096,
+        intermediate_size=11008,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=None,
+        hidden_act="silu",
+        max_position_embeddings=2048,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        pretraining_tp=1,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        rope_scaling=None,
+        attention_bias=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.pretraining_tp = pretraining_tp
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self._rope_scaling_validation()
+        self.attention_bias = attention_bias
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    def _rope_scaling_validation(self):
+        """
+        Validate the `rope_scaling` configuration.
+        """
+        if self.rope_scaling is None:
+            return
+
+        if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
+            raise ValueError(
+                "`rope_scaling` must be a dictionary with with two fields, `type` and `factor`, "
+                f"got {self.rope_scaling}"
+            )
+        rope_scaling_type = self.rope_scaling.get("type", None)
+        rope_scaling_factor = self.rope_scaling.get("factor", None)
+        if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]:
+            raise ValueError(
+                f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
+            )
+        if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
+            raise ValueError(f"`rope_scaling`'s factor field must be a float > 1, got {rope_scaling_factor}")
+
+            
+class MplugOwlVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`MplugOwlVisionModel`]. It is used to instantiate
+    a
+     mPLUG-Owl vision encoder according to the specified arguments, defining the model architecture. Instantiating a
+     configuration defaults will yield a similar configuration to that of the mPLUG-Owl
+     [x-plug/x_plug-llama-7b](https://huggingface.co/x-plug/x_plug-llama-7b) architecture.
+
+     Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+     documentation from [`PretrainedConfig`] for more information.
+
+     Args:
+         hidden_size (`int`, *optional*, defaults to 768):
+             Dimensionality of the encoder layers and the pooler layer.
+         intermediate_size (`int`, *optional*, defaults to 3072):
+             Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+         num_hidden_layers (`int`, *optional*, defaults to 12):
+             Number of hidden layers in the Transformer encoder.
+         num_attention_heads (`int`, *optional*, defaults to 12):
+             Number of attention heads for each attention layer in the Transformer encoder.
+         image_size (`int`, *optional*, defaults to 224):
+             The size (resolution) of each image.
+         patch_size (`int`, *optional*, defaults to 32):
+             The size (resolution) of each patch.
+         hidden_act (`str` or `function`, *optional*, defaults to `"quick_gelu"`):
+             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+             `"relu"`, `"selu"` and `"gelu_new"` ``"quick_gelu"` are supported.
+         layer_norm_eps (`float`, *optional*, defaults to 1e-5):
+             The epsilon used by the layer normalization layers.
+         attention_dropout (`float`, *optional*, defaults to 0.0):
+             The dropout ratio for the attention probabilities.
+         initializer_range (`float`, *optional*, defaults to 0.02):
+             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+         initializer_factor (`float`, *optional*, defaults to 1):
+             A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
+             testing).
+
+
+     ```"""
+
+    model_type = "mplug_owl_vision_model"
+
+    def __init__(
+        self,
+        hidden_size=1024,
+        intermediate_size=4096,
+        projection_dim=768,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        num_channels=3,
+        image_size=448,
+        patch_size=14,
+        hidden_act="quick_gelu",
+        layer_norm_eps=1e-6,
+        attention_dropout=0.0,
+        initializer_range=0.02,
+        initializer_factor=1.0,
+        use_flash_attn=False,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.projection_dim = projection_dim
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.attention_dropout = attention_dropout
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+        self.use_flash_attn = use_flash_attn
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the vision config dict if we are loading from MplugOwlConfig
+        if config_dict.get("model_type") == "mplug-owl":
+            config_dict = config_dict["vision_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+
+class MplugDocOwlHReducerConfig(PretrainedConfig):
+    model_type = "mplug_docowl_hreducer"
+
+    def __init__(
+        self,
+        hidden_size=1024,
+        initializer_range=0.02,
+        layer_norm_eps=1e-6,
+        conv_shape='1x4',
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.hidden_size = hidden_size
+        self.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.conv_shape = conv_shape
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
+
+        # get the visual_abstractor config dict if we are loading from MplugOwlConfig
+        if config_dict.get("model_type") == "mplug-docowl":
+            config_dict = config_dict["hreducer_config"]
+
+        if "model_type" in config_dict and hasattr(cls, "model_type") and config_dict["model_type"] != cls.model_type:
+            logger.warning(
+                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
+                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
+            )
+
+        return cls.from_dict(config_dict, **kwargs)
+
+DEFAULT_VISUAL_CONFIG = {
+    "visual_model": MplugOwlVisionConfig().to_dict(),
+    "visual_hreducer": MplugDocOwlHReducerConfig().to_dict()
+}
+
+class MPLUGDocOwlConfig(LlamaConfig):
+    model_type = "mplug_docowl"
+    def __init__(self, visual_config=None, **kwargs):
+        if visual_config is None:
+            self.visual_config = DEFAULT_VISUAL_CONFIG
+        else:
+            self.visual_config = visual_config
+        
+        super().__init__(
+            **kwargs,
+        )
+        
+if __name__ == "__main__":
+    print(MplugOwlVisionConfig().to_dict())
\ No newline at end of file

mplug_docowl/model/convert_mplug_docowl_weight_to_hf.py

+# Copyright 2023 DAMO Academy and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import argparse
+import gc
+import json
+import math
+import os
+import shutil
+import warnings
+
+import torch
+
+from transformers import LlamaTokenizer
+from .configuration_mplug_docowl import MPLUGDocOwlConfig
+from icecream import ic
+
+try:
+    from transformers import LlamaTokenizerFast
+except ImportError as e:
+    warnings.warn(e)
+    warnings.warn(
+        "The converted tokenizer will be the `slow` tokenizer. To use the fast, update your `tokenizers` library and re-run the tokenizer conversion"
+    )
+    LlamaTokenizerFast = None
+
+"""
+Sample usage:
+
+```
+python3 /pure-mlo-scratch/sfan/model-parallel-trainer/llama2megatron/convert_llama2hf.py \
+    --input_dir /pure-mlo-scratch/llama/ --model_size 7 --output_dir /pure-mlo-scratch/llama/converted_HF_7B
+```
+
+Thereafter, models can be loaded via:
+
+```py
+from transformers import LlamaForCausalLM, LlamaTokenizer
+
+model = LlamaForCausalLM.from_pretrained("/output/path")
+tokenizer = LlamaTokenizer.from_pretrained("/output/path")
+```
+
+Important note: you need to be able to host the whole model in RAM to execute this script (even if the biggest versions
+come in several checkpoints they each contain a part of each weight of the model, so we need to load them all in RAM).
+"""
+
+llama_s2layer = {7: 32, 13: 40, 30: 60, 65: 80, 70: 80}
+llama_s2heads = {7: 32, 13: 40, 30: 52, 65: 64, 70: 64}
+llama_s2dense = {7: 11008, 13: 13824, 30: 17920, 65: 22016,
+                 70: 28672}  # should be (2/3)*4*d, but it isn't exaclty that
+llama_s2hidden = {7: 4096, 13: 5120, 32: 6656, 65: 8192, 70: 8192}
+
+
+def compute_intermediate_size(n):
+    return int(math.ceil(n * 8 / 3) + 255) // 256 * 256
+
+
+def read_json(path):
+    with open(path, "r") as f:
+        return json.load(f)
+
+
+def write_json(text, path):
+    with open(path, "w") as f:
+        json.dump(text, f)
+
+
+def write_model(model_path, 
+                input_base_path, 
+                model_size,
+                num_input_shards=1,
+                num_output_shards=2,
+                skip_permute=True,
+                norm_eps=1e-05):
+    # if os.path.exists(model_path):
+    #     shutil.rmtree(model_path)
+    os.makedirs(model_path, exist_ok=True)
+    # tmp_model_path = os.path.join(model_path, "tmp")
+    tmp_model_path = model_path
+    os.makedirs(tmp_model_path, exist_ok=True)
+
+    num_shards = num_input_shards
+    n_layers = llama_s2layer[model_size]
+    n_heads = llama_s2heads[model_size]
+    n_heads_per_shard = n_heads // num_shards
+    n_dense = llama_s2dense[model_size]
+    n_hidden = llama_s2hidden[model_size]
+    hidden_per_head = n_hidden // n_heads
+    base = 10000.0
+    inv_freq = 1.0 / (base ** (torch.arange(0, hidden_per_head, 2).float() / hidden_per_head))
+
+    # permute for sliced rotary
+    def permute(w, skip_permute=skip_permute):
+        if skip_permute:
+            return w
+        return w.view(n_heads, n_hidden // n_heads // 2, 2, n_hidden).transpose(1, 2).reshape(n_hidden, n_hidden)
+
+    print(f"Fetching all parameters from the checkpoint at {input_base_path}.")
+    # Load weights
+    if num_shards==1:
+        # Not sharded
+        # (The sharded implementation would also work, but this is simpler.)
+        # /pure-mlo-scratch/alhernan/megatron-data/checkpoints/llama2-7b-tp4-pp1-optim/release/mp_rank_00/model_optim_rng.pt
+        if os.path.exists(os.path.join(input_base_path, 'release')):
+            filename = os.path.join(input_base_path, 'release', 'mp_rank_00', 'model_optim_rng.pt')
+        elif input_base_path.split('/')[-1].startswith('iter_'):
+            iteration = eval(input_base_path.split('/')[-1].replace('iter_', '').lstrip('0'))
+            load_dir = '/'.join(input_base_path.split('/')[:-1])
+            filename = os.path.join(input_base_path, 'mp_rank_00', 'model_optim_rng.pt')
+            if not os.path.exists(filename):
+                filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+        else:
+            tracker_filename = os.path.join(input_base_path, 'latest_checkpointed_iteration.txt')
+            with open(tracker_filename, 'r') as f:
+                metastring = f.read().strip()
+            iteration = 'iter_{:07d}'.format(int(metastring))
+            filename = os.path.join(input_base_path, iteration, 'mp_rank_00', 'model_optim_rng.pt')
+        if not os.path.exists(filename):
+            filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+        original_filename = filename
+        loaded = torch.load(filename, map_location="cpu")['model']['language_model']
+
+    else:
+        # Sharded
+        filenames = []
+        for i in range(num_shards):
+            if os.path.exists(os.path.join(input_base_path, 'release')):
+                filename = os.path.join(input_base_path, 'release', f'mp_rank_{i:02d}', 'model_optim_rng.pt')
+            else:
+                tracker_filename = os.path.join(input_base_path, 'latest_checkpointed_iteration.txt')
+                with open(tracker_filename, 'r') as f:
+                    metastring = f.read().strip()
+                iteration = 'iter_{:07d}'.format(int(metastring))
+                filename = os.path.join(input_base_path, iteration, f'mp_rank_{i:02d}', 'model_optim_rng.pt')
+            if not os.path.exists(filename):
+                filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+            filenames.append(filename)
+        loaded = [
+            torch.load(filenames[i], map_location="cpu")['model']['language_model']
+            for i in range(num_shards)
+        ]
+
+    print('Llama-Megatron Loaded!')
+    param_count = 0
+    index_dict = {"weight_map": {}}
+    
+    print(f'Weighted Converting for {n_layers} layers...')
+    for layer_i in range(n_layers):
+        print(layer_i)
+        filename = f"pytorch_model-{layer_i + 1}-of-{n_layers + 1}.bin"
+        if num_shards == 1:
+            # Unsharded
+            state_dict = {
+                f"model.layers.{layer_i}.self_attn.q_proj.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.q_proj.weight"],
+                f"model.layers.{layer_i}.self_attn.k_proj.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.k_proj.multiway.0.weight"],
+                f"model.layers.{layer_i}.self_attn.v_proj.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.v_proj.multiway.0.weight"],
+                f"model.layers.{layer_i}.self_attn.k_proj.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.k_proj.multiway.1.weight"],
+                f"model.layers.{layer_i}.self_attn.v_proj.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.v_proj.multiway.1.weight"],
+                f"model.layers.{layer_i}.self_attn.o_proj.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.o_proj.weight"],
+                f"model.layers.{layer_i}.mlp.gate_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.gate_proj.weight"],
+                f"model.layers.{layer_i}.mlp.down_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.down_proj.weight"],
+                f"model.layers.{layer_i}.mlp.up_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.up_proj.weight"],
+                f"model.layers.{layer_i}.input_layernorm.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.input_layernorm.multiway.0.weight"],
+                f"model.layers.{layer_i}.post_attention_layernorm.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.post_attention_layernorm.multiway.0.weight"],
+                f"model.layers.{layer_i}.input_layernorm.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.input_layernorm.multiway.1.weight"],
+                f"model.layers.{layer_i}.post_attention_layernorm.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.post_attention_layernorm.multiway.1.weight"],
+            }
+        else:
+            raise NotImplemented
+
+        state_dict[f"model.layers.{layer_i}.self_attn.rotary_emb.inv_freq"] = inv_freq
+        for k, v in state_dict.items():
+            index_dict["weight_map"][k] = filename
+            param_count += v.numel()
+        torch.save(state_dict, os.path.join(tmp_model_path, filename))
+        print(f'Sharded file saved to {filename}')
+
+    filename = f"pytorch_model-{n_layers + 1}-of-{n_layers + 1}.bin"
+    if num_shards==1:
+        # Unsharded
+        state_dict = {
+            "model.embed_tokens.weight": loaded['embedding']['word_embeddings']['weight'],
+            "model.norm.weight": loaded['encoder']['norm.weight'],
+            "lm_head.weight": loaded['encoder']['lm_head.weight'],
+        }
+    else:
+        state_dict = {
+            "model.embed_tokens.weight": loaded[0]['embedding']['word_embeddings']['weight'],
+            "model.norm.weight": loaded[0]['encoder']['norm.weight'],
+            "lm_head.weight": loaded[0]['encoder']['lm_head.weight'],
+        }
+        
+    
+    loaded_all = torch.load(original_filename, map_location="cpu")['model']
+    # Vision Part
+    state_dict.update({
+        "model.vision_model.embeddings.cls_token": loaded_all['vision_model']['cls_token'],
+        "model.vision_model.embeddings.patch_embed.weight": loaded_all['vision_model']['patch_embed']['weight'],
+        "model.vision_model.embeddings.position_embedding": loaded_all['vision_model']['position_embeddings'],
+        "model.vision_model.embeddings.pre_layernorm.bias": loaded_all['vision_model']['pre_layernorm']['bias'],
+        "model.vision_model.embeddings.pre_layernorm.weight": loaded_all['vision_model']['pre_layernorm']['weight'],
+        "model.vision_model.post_layernorm.bias": loaded_all['vision_model']['transformer']['final_layernorm.bias'],
+        "model.vision_model.post_layernorm.weight": loaded_all['vision_model']['transformer']['final_layernorm.weight'],
+    })
+    for v_layer_idx in range(24):
+        state_dict.update({
+            f"model.vision_model.encoder.layers.{v_layer_idx}.input_layernorm.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.input_layernorm.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.input_layernorm.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.input_layernorm.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc1.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_h_to_4h.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc1.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_h_to_4h.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc2.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_4h_to_h.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc2.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_4h_to_h.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.post_attention_layernorm.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.post_attention_layernorm.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.post_attention_layernorm.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.post_attention_layernorm.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.dense.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.dense.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.dense.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.dense.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.query_key_value.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.query_key_value.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.query_key_value.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.query_key_value.weight'],
+        })
+
+    # Vision2Text Part: HReducer 
+    state_dict.update({
+        "model.vision2text.ln_q.weight": loaded_all['hreducer3']['ln_q']['weight'],
+        "model.vision2text.ln_q.bias": loaded_all['hreducer3']['ln_q']['bias'],
+        "model.vision2text.visual_fc.bias": loaded_all['hreducer3']['visual_fc']['bias'],
+        "model.vision2text.visual_fc.weight": loaded_all['hreducer3']['visual_fc']['weight'],
+        "model.vision2text.vit_eos": loaded_all['hreducer3']['vit_eos'],  
+    })
+    # reducer_before conv (layer 0) + gleu  (layer 1)
+    state_dict.update({
+        f"model.vision2text.reducer_before.0.weight": loaded_all['hreducer3']['reducer_before']["0.weight"],
+        f"model.vision2text.reducer_before.0.bias": loaded_all['hreducer3']['reducer_before']["0.bias"],
+    })
+    # reducer conv 
+    state_dict.update({
+        f"model.vision2text.reducer.weight": loaded_all['hreducer3']['reducer']["weight"],
+        f"model.vision2text.reducer.bias": loaded_all['hreducer3']['reducer']["bias"],
+    })
+
+    for k, v in state_dict.items():
+        # ic(k, v)
+        index_dict["weight_map"][k] = filename
+        param_count += v.numel()
+    torch.save(state_dict, os.path.join(tmp_model_path, filename))
+
+    # Write configs
+    index_dict["metadata"] = {"total_size": param_count * 2}
+    write_json(index_dict, os.path.join(tmp_model_path, "pytorch_model.bin.index.json"))
+
+    config = MPLUGDocOwlConfig()
+    config.save_pretrained(tmp_model_path)
+
+    # Make space so we can load the model properly now.
+    del state_dict
+    del loaded
+    del loaded_all
+    gc.collect()
+
+def write_tokenizer(tokenizer_path, input_tokenizer_path):
+    # Initialize the tokenizer based on the `spm` model
+    tokenizer_class = LlamaTokenizer if LlamaTokenizerFast is None else LlamaTokenizerFast
+    print(f"Saving a {tokenizer_class.__name__} to {tokenizer_path}.")
+    tokenizer = tokenizer_class(input_tokenizer_path)
+    tokenizer.save_pretrained(tokenizer_path)
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--input_dir",
+        help="Location of LLaMA_Megatron weights",
+    )
+    parser.add_argument(
+        "--model_size",
+        type=int,
+        default=7,
+        choices=[7, 13, 30, 65, 70],
+    )
+    parser.add_argument(
+        "--num_input_shards",
+        type=int,
+        default=1,
+    )
+    parser.add_argument(
+        "--num_output_shards",
+        type=int,
+        default=1,
+    )
+    parser.add_argument('--skip_permute', action='store_true')
+    
+    parser.add_argument(
+        "--output_dir",
+        help="Location to write HF model and tokenizer",
+    )
+    
+    args = parser.parse_args()
+    write_model(
+        model_path=args.output_dir,
+        input_base_path=args.input_dir,
+        model_size=args.model_size,
+        num_input_shards=args.num_input_shards,
+        num_output_shards=args.num_output_shards,
+        skip_permute=args.skip_permute
+    )
+    
+
+if __name__ == "__main__":
+    main()

mplug_docowl/model/convert_mplug_docowl_weight_to_hf_v2.py

+# Copyright 2023 DAMO Academy and The HuggingFace Inc. team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import argparse
+import gc
+import json
+import math
+import os
+import shutil
+import warnings
+
+import torch
+
+from transformers import LlamaTokenizer
+from .configuration_mplug_docowl import MPLUGDocOwlConfig
+from icecream import ic
+
+try:
+    from transformers import LlamaTokenizerFast
+except ImportError as e:
+    warnings.warn(e)
+    warnings.warn(
+        "The converted tokenizer will be the `slow` tokenizer. To use the fast, update your `tokenizers` library and re-run the tokenizer conversion"
+    )
+    LlamaTokenizerFast = None
+
+"""
+Sample usage:
+
+```
+python3 /pure-mlo-scratch/sfan/model-parallel-trainer/llama2megatron/convert_llama2hf.py \
+    --input_dir /pure-mlo-scratch/llama/ --model_size 7 --output_dir /pure-mlo-scratch/llama/converted_HF_7B
+```
+
+Thereafter, models can be loaded via:
+
+```py
+from transformers import LlamaForCausalLM, LlamaTokenizer
+
+model = LlamaForCausalLM.from_pretrained("/output/path")
+tokenizer = LlamaTokenizer.from_pretrained("/output/path")
+```
+
+Important note: you need to be able to host the whole model in RAM to execute this script (even if the biggest versions
+come in several checkpoints they each contain a part of each weight of the model, so we need to load them all in RAM).
+"""
+
+llama_s2layer = {7: 32, 13: 40, 30: 60, 65: 80, 70: 80}
+llama_s2heads = {7: 32, 13: 40, 30: 52, 65: 64, 70: 64}
+llama_s2dense = {7: 11008, 13: 13824, 30: 17920, 65: 22016,
+                 70: 28672}  # should be (2/3)*4*d, but it isn't exaclty that
+llama_s2hidden = {7: 4096, 13: 5120, 32: 6656, 65: 8192, 70: 8192}
+
+
+def compute_intermediate_size(n):
+    return int(math.ceil(n * 8 / 3) + 255) // 256 * 256
+
+
+def read_json(path):
+    with open(path, "r") as f:
+        return json.load(f)
+
+
+def write_json(text, path):
+    with open(path, "w") as f:
+        json.dump(text, f)
+
+
+def write_model(model_path, 
+                input_base_path, 
+                model_size,
+                num_input_shards=1,
+                num_output_shards=2,
+                skip_permute=True,
+                norm_eps=1e-05):
+    # if os.path.exists(model_path):
+    #     shutil.rmtree(model_path)
+    os.makedirs(model_path, exist_ok=True)
+    # tmp_model_path = os.path.join(model_path, "tmp")
+    tmp_model_path = model_path
+    os.makedirs(tmp_model_path, exist_ok=True)
+
+    num_shards = num_input_shards
+    n_layers = llama_s2layer[model_size]
+    n_heads = llama_s2heads[model_size]
+    n_heads_per_shard = n_heads // num_shards
+    n_dense = llama_s2dense[model_size]
+    n_hidden = llama_s2hidden[model_size]
+    hidden_per_head = n_hidden // n_heads
+    base = 10000.0
+    inv_freq = 1.0 / (base ** (torch.arange(0, hidden_per_head, 2).float() / hidden_per_head))
+
+    # permute for sliced rotary
+    def permute(w, skip_permute=skip_permute):
+        if skip_permute:
+            return w
+        return w.view(n_heads, n_hidden // n_heads // 2, 2, n_hidden).transpose(1, 2).reshape(n_hidden, n_hidden)
+
+    print(f"Fetching all parameters from the checkpoint at {input_base_path}.")
+    # Load weights
+    if num_shards==1:
+        # Not sharded
+        # (The sharded implementation would also work, but this is simpler.)
+        # /pure-mlo-scratch/alhernan/megatron-data/checkpoints/llama2-7b-tp4-pp1-optim/release/mp_rank_00/model_optim_rng.pt
+        if os.path.exists(os.path.join(input_base_path, 'release')):
+            filename = os.path.join(input_base_path, 'release', 'mp_rank_00', 'model_optim_rng.pt')
+        elif input_base_path.split('/')[-1].startswith('iter_'):
+            iteration = eval(input_base_path.split('/')[-1].replace('iter_', '').lstrip('0'))
+            load_dir = '/'.join(input_base_path.split('/')[:-1])
+            filename = os.path.join(input_base_path, 'mp_rank_00', 'model_optim_rng.pt')
+            if not os.path.exists(filename):
+                filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+        else:
+            tracker_filename = os.path.join(input_base_path, 'latest_checkpointed_iteration.txt')
+            with open(tracker_filename, 'r') as f:
+                metastring = f.read().strip()
+            iteration = 'iter_{:07d}'.format(int(metastring))
+            filename = os.path.join(input_base_path, iteration, 'mp_rank_00', 'model_optim_rng.pt')
+        if not os.path.exists(filename):
+            filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+        original_filename = filename
+        loaded = torch.load(filename, map_location="cpu")['model']['language_model']
+
+    else:
+        # Sharded
+        filenames = []
+        for i in range(num_shards):
+            if os.path.exists(os.path.join(input_base_path, 'release')):
+                filename = os.path.join(input_base_path, 'release', f'mp_rank_{i:02d}', 'model_optim_rng.pt')
+            else:
+                tracker_filename = os.path.join(input_base_path, 'latest_checkpointed_iteration.txt')
+                with open(tracker_filename, 'r') as f:
+                    metastring = f.read().strip()
+                iteration = 'iter_{:07d}'.format(int(metastring))
+                filename = os.path.join(input_base_path, iteration, f'mp_rank_{i:02d}', 'model_optim_rng.pt')
+            if not os.path.exists(filename):
+                filename = filename.replace('model_optim_rng.pt', 'model_rng.pt')
+            filenames.append(filename)
+        loaded = [
+            torch.load(filenames[i], map_location="cpu")['model']['language_model']
+            for i in range(num_shards)
+        ]
+
+    print('Llama-Megatron Loaded!')
+    param_count = 0
+    index_dict = {"weight_map": {}}
+    state_dict = {}
+    print(f'Weighted Converting for {n_layers} layers...')
+    for layer_i in range(n_layers):
+        print(layer_i)
+        # filename = f"pytorch_model-{layer_i + 1}-of-{n_layers + 1}.bin"
+        if num_shards == 1:
+            # Unsharded
+            state_dict.update({
+                f"model.layers.{layer_i}.self_attn.q_proj.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.q_proj.weight"],
+                f"model.layers.{layer_i}.self_attn.k_proj.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.k_proj.multiway.0.weight"],
+                f"model.layers.{layer_i}.self_attn.v_proj.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.v_proj.multiway.0.weight"],
+                f"model.layers.{layer_i}.self_attn.k_proj.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.k_proj.multiway.1.weight"],
+                f"model.layers.{layer_i}.self_attn.v_proj.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.v_proj.multiway.1.weight"],
+                f"model.layers.{layer_i}.self_attn.o_proj.weight": loaded['encoder'][f"layers.{layer_i}.self_attention.o_proj.weight"],
+                f"model.layers.{layer_i}.mlp.gate_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.gate_proj.weight"],
+                f"model.layers.{layer_i}.mlp.down_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.down_proj.weight"],
+                f"model.layers.{layer_i}.mlp.up_proj.weight": loaded['encoder'][f"layers.{layer_i}.mlp.up_proj.weight"],
+                f"model.layers.{layer_i}.input_layernorm.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.input_layernorm.multiway.0.weight"],
+                f"model.layers.{layer_i}.post_attention_layernorm.multiway.0.weight": loaded['encoder'][f"layers.{layer_i}.post_attention_layernorm.multiway.0.weight"],
+                f"model.layers.{layer_i}.input_layernorm.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.input_layernorm.multiway.1.weight"],
+                f"model.layers.{layer_i}.post_attention_layernorm.multiway.1.weight": loaded['encoder'][f"layers.{layer_i}.post_attention_layernorm.multiway.1.weight"],
+            })
+        else:
+            raise NotImplemented
+
+        state_dict[f"model.layers.{layer_i}.self_attn.rotary_emb.inv_freq"] = inv_freq
+        for k, v in state_dict.items():
+            index_dict["weight_map"][k] = filename
+            param_count += v.numel()
+        # torch.save(state_dict, os.path.join(tmp_model_path, filename))
+        # print(f'Sharded file saved to {filename}')
+
+    # filename = f"pytorch_model-{n_layers + 1}-of-{n_layers + 1}.bin"
+    filename = "pytorch_model.bin"
+    if num_shards==1:
+        # Unsharded
+        state_dict.update({
+            "model.embed_tokens.weight": loaded['embedding']['word_embeddings']['weight'],
+            "model.norm.weight": loaded['encoder']['norm.weight'],
+            "lm_head.weight": loaded['encoder']['lm_head.weight'],
+        })
+    else:
+        state_dict.update({
+            "model.embed_tokens.weight": loaded[0]['embedding']['word_embeddings']['weight'],
+            "model.norm.weight": loaded[0]['encoder']['norm.weight'],
+            "lm_head.weight": loaded[0]['encoder']['lm_head.weight'],
+        })
+        
+    loaded_all = torch.load(original_filename, map_location="cpu")['model']
+    # Vision Part
+    state_dict.update({
+        "model.vision_model.embeddings.cls_token": loaded_all['vision_model']['cls_token'],
+        "model.vision_model.embeddings.patch_embed.weight": loaded_all['vision_model']['patch_embed']['weight'],
+        "model.vision_model.embeddings.position_embedding": loaded_all['vision_model']['position_embeddings'],
+        "model.vision_model.embeddings.pre_layernorm.bias": loaded_all['vision_model']['pre_layernorm']['bias'],
+        "model.vision_model.embeddings.pre_layernorm.weight": loaded_all['vision_model']['pre_layernorm']['weight'],
+        "model.vision_model.post_layernorm.bias": loaded_all['vision_model']['transformer']['final_layernorm.bias'],
+        "model.vision_model.post_layernorm.weight": loaded_all['vision_model']['transformer']['final_layernorm.weight'],
+    })
+    for v_layer_idx in range(24):
+        state_dict.update({
+            f"model.vision_model.encoder.layers.{v_layer_idx}.input_layernorm.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.input_layernorm.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.input_layernorm.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.input_layernorm.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc1.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_h_to_4h.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc1.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_h_to_4h.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc2.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_4h_to_h.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.mlp.fc2.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.mlp.dense_4h_to_h.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.post_attention_layernorm.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.post_attention_layernorm.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.post_attention_layernorm.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.post_attention_layernorm.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.dense.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.dense.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.dense.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.dense.weight'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.query_key_value.bias": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.query_key_value.bias'],
+            f"model.vision_model.encoder.layers.{v_layer_idx}.self_attn.query_key_value.weight": loaded_all['vision_model']['transformer'][f'layers.{v_layer_idx}.self_attention.query_key_value.weight'],
+        })
+
+    # Vision2Text Part: HReducer 
+    state_dict.update({
+        "model.vision2text.ln_q.weight": loaded_all['hreducer3']['ln_q']['weight'],
+        "model.vision2text.ln_q.bias": loaded_all['hreducer3']['ln_q']['bias'],
+        "model.vision2text.visual_fc.bias": loaded_all['hreducer3']['visual_fc']['bias'],
+        "model.vision2text.visual_fc.weight": loaded_all['hreducer3']['visual_fc']['weight'],
+        "model.vision2text.vit_eos": loaded_all['hreducer3']['vit_eos'],  
+    })
+    # reducer_before conv (layer 0) + gleu  (layer 1)
+    state_dict.update({
+        f"model.vision2text.reducer_before.0.weight": loaded_all['hreducer3']['reducer_before']["0.weight"],
+        f"model.vision2text.reducer_before.0.bias": loaded_all['hreducer3']['reducer_before']["0.bias"],
+    })
+    # reducer conv 
+    state_dict.update({
+        f"model.vision2text.reducer.weight": loaded_all['hreducer3']['reducer']["weight"],
+        f"model.vision2text.reducer.bias": loaded_all['hreducer3']['reducer']["bias"],
+    })
+
+    for k, v in state_dict.items():
+        # ic(k, v)
+        index_dict["weight_map"][k] = filename
+        param_count += v.numel()
+    torch.save(state_dict, os.path.join(tmp_model_path, filename))
+    print(f'save to {os.path.join(tmp_model_path, filename)}')
+
+    # Write configs
+    index_dict["metadata"] = {"total_size": param_count * 2}
+    write_json(index_dict, os.path.join(tmp_model_path, "pytorch_model.bin.index.json"))
+
+    config = MPLUGDocOwlConfig()
+    config.save_pretrained(tmp_model_path)
+
+    # Make space so we can load the model properly now.
+    del state_dict
+    del loaded
+    del loaded_all
+    gc.collect()
+
+def write_tokenizer(tokenizer_path, input_tokenizer_path):
+    # Initialize the tokenizer based on the `spm` model
+    tokenizer_class = LlamaTokenizer if LlamaTokenizerFast is None else LlamaTokenizerFast
+    print(f"Saving a {tokenizer_class.__name__} to {tokenizer_path}.")
+    tokenizer = tokenizer_class(input_tokenizer_path)
+    tokenizer.save_pretrained(tokenizer_path)
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--input_dir",
+        help="Location of LLaMA_Megatron weights",
+    )
+    parser.add_argument(
+        "--model_size",
+        type=int,
+        default=7,
+        choices=[7, 13, 30, 65, 70],
+    )
+    parser.add_argument(
+        "--num_input_shards",
+        type=int,
+        default=1,
+    )
+    parser.add_argument(
+        "--num_output_shards",
+        type=int,
+        default=1,
+    )
+    parser.add_argument('--skip_permute', action='store_true')
+    
+    parser.add_argument(
+        "--output_dir",
+        help="Location to write HF model and tokenizer",
+    )
+    
+    args = parser.parse_args()
+    write_model(
+        model_path=args.output_dir,
+        input_base_path=args.input_dir,
+        model_size=args.model_size,
+        num_input_shards=args.num_input_shards,
+        num_output_shards=args.num_output_shards,
+        skip_permute=args.skip_permute
+    )
+    
+
+if __name__ == "__main__":
+    main()

mplug_docowl/model/modeling_attn_mask_utils.py

+# Copyright 2023 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import List, Optional, Tuple, Union
+
+import torch
+
+
+class AttentionMaskConverter:
+    """
+    A utility attention mask class that allows one to:
+        - Create a causal 4d mask
+        - Create a causal 4d mask with slided window
+        - Convert a 2d attention mask (batch_size, query_length) to a 4d attention mask (batch_size, 1, query_length,
+          key_value_length) that can be multiplied with attention scores
+
+    Parameters:
+        is_causal (`bool`):
+            Whether the attention mask should be a uni-directional (causal) or bi-directional mask.
+
+        sliding_window (`int`, *optional*):
+            Optionally, the sliding window masks can be created if `sliding_window` is defined to a positive integer.
+    """
+
+    def __init__(self, is_causal: bool, sliding_window: Optional[int] = None):
+        self.is_causal = is_causal
+        self.sliding_window = sliding_window
+
+        if self.sliding_window is not None and self.sliding_window <= 0:
+            raise ValueError(
+                f"Make sure that when passing `sliding_window` that its value is a strictly positive integer, not `{self.sliding_window}`"
+            )
+
+    def to_causal_4d(
+        self,
+        batch_size: int,
+        query_length: int,
+        key_value_length: int,
+        dtype: torch.dtype = torch.float32,
+        device: Union[torch.device, "str"] = "cpu",
+    ) -> torch.Tensor:
+        """
+        Creates a causal 4D mask of (bsz, head_dim=1, query_length, key_value_length) shape and adds large negative
+        bias to upper right hand triangular matrix (causal mask).
+        """
+        if not self.is_causal:
+            raise ValueError(f"Please use `to_causal_4d` only if {self.__class__} has `is_causal` set to True.")
+
+        # If shape is not cached, create a new causal mask and cache it
+        input_shape = (batch_size, query_length)
+        past_key_values_length = key_value_length - query_length
+
+        # create causal mask
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        causal_4d_mask = None
+        if input_shape[-1] > 1 or self.sliding_window is not None:
+            causal_4d_mask = self._make_causal_mask(
+                input_shape,
+                dtype,
+                device=device,
+                past_key_values_length=past_key_values_length,
+                sliding_window=self.sliding_window,
+            )
+
+        return causal_4d_mask
+
+    def to_4d(
+        self,
+        attention_mask_2d: torch.Tensor,
+        query_length: int,
+        key_value_length: Optional[int] = None,
+        dtype: torch.dtype = torch.float32,
+    ) -> torch.Tensor:
+        """
+        Converts 2D attention mask to 4D attention mask by expanding mask to (bsz, head_dim=1, query_length,
+        key_value_length) shape and by adding a large negative bias to not-attended positions. If attention_mask is
+        causal, a causal mask will be added.
+        """
+        input_shape = (attention_mask_2d.shape[0], query_length)
+
+        # create causal mask
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        causal_4d_mask = None
+        if (input_shape[-1] > 1 or self.sliding_window is not None) and self.is_causal:
+            if key_value_length is None:
+                raise ValueError(
+                    "This attention mask converter is causal. Make sure to pass `key_value_length` to correctly create a causal mask."
+                )
+
+            past_key_values_length = key_value_length - query_length
+            causal_4d_mask = self._make_causal_mask(
+                input_shape,
+                dtype,
+                device=attention_mask_2d.device,
+                past_key_values_length=past_key_values_length,
+                sliding_window=self.sliding_window,
+            )
+        elif self.sliding_window is not None:
+            raise NotImplementedError("Sliding window is currently only implemented for causal masking")
+
+        # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
+        expanded_attn_mask = self._expand_mask(attention_mask_2d, dtype, tgt_len=input_shape[-1]).to(
+            attention_mask_2d.device
+        )
+        expanded_4d_mask = expanded_attn_mask if causal_4d_mask is None else expanded_attn_mask + causal_4d_mask
+
+        return expanded_4d_mask
+
+    @staticmethod
+    def _make_causal_mask(
+        input_ids_shape: torch.Size,
+        dtype: torch.dtype,
+        device: torch.device,
+        past_key_values_length: int = 0,
+        sliding_window: Optional[int] = None,
+    ):
+        """
+        Make causal mask used for bi-directional self-attention.
+        """
+        bsz, tgt_len = input_ids_shape
+        mask = torch.full((tgt_len, tgt_len), torch.finfo(dtype).min, device=device)
+        mask_cond = torch.arange(mask.size(-1), device=device)
+        mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
+
+        mask = mask.to(dtype)
+
+        if past_key_values_length > 0:
+            mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
+
+        # add lower triangular sliding window mask if necessary
+        if sliding_window is not None:
+            diagonal = past_key_values_length - sliding_window + 1
+
+            context_mask = 1 - torch.triu(torch.ones_like(mask, dtype=torch.int), diagonal=diagonal)
+            mask.masked_fill_(context_mask.bool(), torch.finfo(dtype).min)
+
+        return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
+
+    @staticmethod
+    def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+        """
+        Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
+        """
+        bsz, src_len = mask.size()
+        tgt_len = tgt_len if tgt_len is not None else src_len
+
+        expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
+
+        inverted_mask = 1.0 - expanded_mask
+
+        return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
+
+
+def _prepare_4d_causal_attention_mask(
+    attention_mask: Optional[torch.Tensor],
+    input_shape: Union[torch.Size, Tuple, List],
+    inputs_embeds: torch.Tensor,
+    past_key_values_length: int,
+    sliding_window: Optional[int] = None,
+):
+    """
+    Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
+    `(batch_size, key_value_length)`
+
+    Args:
+        attention_mask (`torch.Tensor` or `None`):
+            A 2D attention mask of shape `(batch_size, key_value_length)`
+        input_shape (`tuple(int)` or `list(int)` or `torch.Size`):
+            The input shape should be a tuple that defines `(batch_size, query_length)`.
+        inputs_embeds (`torch.Tensor`):
+            The embedded inputs as a torch Tensor.
+        past_key_values_length (`int`):
+            The length of the key value cache.
+        sliding_window (`int`, *optional*):
+            If the model uses windowed attention, a sliding window should be passed.
+    """
+    attn_mask_converter = AttentionMaskConverter(is_causal=True, sliding_window=sliding_window)
+
+    key_value_length = input_shape[-1] + past_key_values_length
+
+    # 4d mask is passed through the layers
+    if attention_mask is not None:
+        attention_mask = attn_mask_converter.to_4d(
+            attention_mask, input_shape[-1], key_value_length, dtype=inputs_embeds.dtype
+        )
+    else:
+        attention_mask = attn_mask_converter.to_causal_4d(
+            input_shape[0], input_shape[-1], key_value_length, dtype=inputs_embeds.dtype, device=inputs_embeds.device
+        )
+
+    return attention_mask
+
+
+def _prepare_4d_attention_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+    """
+    Creates a non-causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape
+    `(batch_size, key_value_length)`
+
+    Args:
+        mask (`torch.Tensor` or `None`):
+            A 2D attention mask of shape `(batch_size, key_value_length)`
+        dtype (`torch.dtype`):
+            The torch dtype the created mask shall have.
+        tgt_len (`int`):
+            The target length or query length the created mask shall have.
+    """
+    return AttentionMaskConverter._expand_mask(mask=mask, dtype=dtype, tgt_len=tgt_len)
+
+
+def _create_4d_causal_attention_mask(
+    input_shape: Union[torch.Size, Tuple, List],
+    dtype: torch.dtype,
+    device: torch.device,
+    past_key_values_length: int = 0,
+    sliding_window: Optional[int] = None,
+):
+    """
+    Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)`
+
+    Args:
+        input_shape (`tuple(int)` or `list(int)` or `torch.Size`):
+            The input shape should be a tuple that defines `(batch_size, query_length)`.
+        dtype (`torch.dtype`):
+            The torch dtype the created mask shall have.
+        device (`int`):
+            The torch device the created mask shall have.
+        sliding_window (`int`, *optional*):
+            If the model uses windowed attention, a sliding window should be passed.
+    """
+    attn_mask_converter = AttentionMaskConverter(is_causal=True, sliding_window=sliding_window)
+
+    key_value_length = past_key_values_length + input_shape[-1]
+    attention_mask = attn_mask_converter.to_causal_4d(
+        input_shape[0], input_shape[-1], key_value_length, dtype=dtype, device=device
+    )
+
+    return attention_mask
\ No newline at end of file

mplug_docowl/model/modeling_llama2.py

+import math
+import warnings
+from functools import partial
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from torch import nn
+
+import transformers
+from transformers.models.llama.modeling_llama import *
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+from .modeling_attn_mask_utils import _prepare_4d_causal_attention_mask
+from .configuration_mplug_docowl import LlamaConfig
+
+class MultiwayNetwork(nn.Module):
+
+    def __init__(self, module_provider, num_multiway=2):
+        super(MultiwayNetwork, self).__init__()
+
+        self.multiway = torch.nn.ModuleList([module_provider() for _ in range(num_multiway)])
+    
+    def forward(self, hidden_states, multiway_indices):
+
+        if len(self.multiway) == 1:
+            return self.multiway[0](hidden_states)
+
+        output_hidden_states = torch.empty_like(hidden_states)
+        
+        for idx, subway in enumerate(self.multiway):
+            local_indices = multiway_indices.eq(idx).nonzero(as_tuple=True)
+            hidden = hidden_states[local_indices].unsqueeze(1).contiguous()
+            if hidden.numel():
+                output = subway(hidden)
+                if isinstance(output, tuple):
+                    output = output[0]
+                output = output.squeeze(1)
+                output_hidden_states[local_indices] = output
+        
+        return output_hidden_states.contiguous()
+    
+
+class LlamaAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config: LlamaConfig):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.num_key_value_heads = config.num_key_value_heads
+        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
+        self.max_position_embeddings = config.max_position_embeddings
+        self.rope_theta = config.rope_theta
+
+        if (self.head_dim * self.num_heads) != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
+                f" and `num_heads`: {self.num_heads})."
+            )
+        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=config.attention_bias)
+        self.k_proj = MultiwayNetwork(module_provider=partial(
+            nn.Linear, in_features=self.hidden_size, out_features=self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        )
+        self.v_proj = MultiwayNetwork(module_provider=partial(
+            nn.Linear, in_features=self.hidden_size, out_features=self.num_key_value_heads * self.head_dim, bias=config.attention_bias)
+        )
+        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=config.attention_bias)
+        self._init_rope()
+
+    def _init_rope(self):
+        if self.config.rope_scaling is None:
+            self.rotary_emb = LlamaRotaryEmbedding(
+                self.head_dim,
+                max_position_embeddings=self.max_position_embeddings,
+                base=self.rope_theta,
+            )
+        else:
+            scaling_type = self.config.rope_scaling["type"]
+            scaling_factor = self.config.rope_scaling["factor"]
+            if scaling_type == "linear":
+                self.rotary_emb = LlamaLinearScalingRotaryEmbedding(
+                    self.head_dim,
+                    max_position_embeddings=self.max_position_embeddings,
+                    scaling_factor=scaling_factor,
+                    base=self.rope_theta,
+                )
+            elif scaling_type == "dynamic":
+                self.rotary_emb = LlamaDynamicNTKScalingRotaryEmbedding(
+                    self.head_dim,
+                    max_position_embeddings=self.max_position_embeddings,
+                    scaling_factor=scaling_factor,
+                    base=self.rope_theta,
+                )
+            else:
+                raise ValueError(f"Unknown RoPE scaling type {scaling_type}")
+
+    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
+        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        padding_mask: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states, )
+        key_states = self.k_proj(hidden_states, modality_indicators)
+        value_states = self.v_proj(hidden_states, modality_indicators)
+
+        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+
+        kv_seq_len = key_states.shape[-2]
+        if past_key_value is not None:
+            kv_seq_len += past_key_value[0].shape[-2]
+        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
+
+        if past_key_value is not None:
+            # reuse k, v, self_attention
+            key_states = torch.cat([past_key_value[0], key_states], dim=2)
+            value_states = torch.cat([past_key_value[1], value_states], dim=2)
+
+        past_key_value = (key_states, value_states) if use_cache else None
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query_states.dtype)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2).contiguous()
+
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+
+class LlamaDecoderLayer(nn.Module):
+    def __init__(self, config: LlamaConfig):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = LlamaAttention(config=config)
+        self.mlp = LlamaMLP(config)
+        self.input_layernorm = MultiwayNetwork(module_provider=partial(
+            LlamaRMSNorm, hidden_size=config.hidden_size, eps=config.rms_norm_eps
+        ))
+        self.post_attention_layernorm = MultiwayNetwork(module_provider=partial(
+            LlamaRMSNorm, hidden_size=config.hidden_size, eps=config.rms_norm_eps
+        ))
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        modality_indicators: torch.Tensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Tuple[torch.Tensor]] = None,
+        output_attentions: Optional[bool] = False,
+        use_cache: Optional[bool] = False,
+    ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            use_cache (`bool`, *optional*):
+                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
+                (see `past_key_values`).
+            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
+        """
+
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states, modality_indicators)
+
+        # Self Attention
+        hidden_states, self_attn_weights, present_key_value = self.self_attn(
+            hidden_states=hidden_states,
+            modality_indicators=modality_indicators,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+        )
+        hidden_states = residual + hidden_states
+
+        # Fully Connected
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states, modality_indicators)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (self_attn_weights,)
+
+        if use_cache:
+            outputs += (present_key_value,)
+
+        return outputs
+
+
+def model_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    modality_indicators: torch.Tensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+) -> Union[Tuple, BaseModelOutputWithPast]:
+    output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+    output_hidden_states = (
+        output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+    )
+    use_cache = use_cache if use_cache is not None else self.config.use_cache
+
+    return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+    # retrieve input_ids and inputs_embeds
+    if input_ids is not None and inputs_embeds is not None:
+        raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
+    elif input_ids is not None:
+        batch_size, seq_length = input_ids.shape
+    elif inputs_embeds is not None:
+        batch_size, seq_length, _ = inputs_embeds.shape
+    else:
+        raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")
+
+    seq_length_with_past = seq_length
+    past_key_values_length = 0
+
+    if past_key_values is not None:
+        past_key_values_length = past_key_values[0][0].shape[2]
+        seq_length_with_past = seq_length_with_past + past_key_values_length
+
+    if position_ids is None:
+        device = input_ids.device if input_ids is not None else inputs_embeds.device
+        position_ids = torch.arange(
+            past_key_values_length, seq_length + past_key_values_length, dtype=torch.long, device=device
+        )
+        position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
+    else:
+        position_ids = position_ids.view(-1, seq_length).long()
+
+    if inputs_embeds is None:
+        inputs_embeds = self.embed_tokens(input_ids)
+    # embed positions
+    if attention_mask is None:
+        attention_mask = torch.ones(
+            (batch_size, seq_length_with_past), dtype=torch.bool, device=inputs_embeds.device
+        )
+    attention_mask = self._prepare_decoder_attention_mask(
+        attention_mask, (batch_size, seq_length), inputs_embeds, past_key_values_length
+    )
+
+    hidden_states = inputs_embeds
+
+    if self.gradient_checkpointing and self.training:
+        if use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+            )
+            use_cache = False
+
+    # decoder layers
+    all_hidden_states = () if output_hidden_states else None
+    all_self_attns = () if output_attentions else None
+    next_decoder_cache = () if use_cache else None
+
+    for idx, decoder_layer in enumerate(self.layers):
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        past_key_value = past_key_values[idx] if past_key_values is not None else None
+
+        if self.gradient_checkpointing and self.training:
+
+            def create_custom_forward(module):
+                def custom_forward(*inputs):
+                    # None for past_key_value
+                    return module(*inputs, past_key_value, output_attentions)
+
+                return custom_forward
+
+            layer_outputs = torch.utils.checkpoint.checkpoint(
+                create_custom_forward(decoder_layer),
+                hidden_states,
+                modality_indicators,
+                attention_mask,
+                position_ids,
+            )
+        else:
+            layer_outputs = decoder_layer(
+                hidden_states,
+                modality_indicators=modality_indicators,
+                attention_mask=attention_mask,
+                position_ids=position_ids,
+                past_key_value=past_key_value,
+                output_attentions=output_attentions,
+                use_cache=use_cache,
+            )
+
+        hidden_states = layer_outputs[0]
+
+        if use_cache:
+            next_decoder_cache += (layer_outputs[2 if output_attentions else 1],)
+
+        if output_attentions:
+            all_self_attns += (layer_outputs[1],)
+
+    hidden_states = self.norm(hidden_states)
+
+    # add hidden states from the last decoder layer
+    if output_hidden_states:
+        all_hidden_states += (hidden_states,)
+
+    next_cache = next_decoder_cache if use_cache else None
+    if not return_dict:
+        return tuple(v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns] if v is not None)
+    return BaseModelOutputWithPast(
+        last_hidden_state=hidden_states,
+        past_key_values=next_cache,
+        hidden_states=all_hidden_states,
+        attentions=all_self_attns,
+    )
+
+
+def causal_model_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    modality_indicators: torch.Tensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    labels: Optional[torch.LongTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+) -> Union[Tuple, CausalLMOutputWithPast]:
+    r"""
+    Args:
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+    Returns:
+
+    Example:
+
+    ```python
+    >>> from transformers import AutoTokenizer, LlamaForCausalLM
+
+    >>> model = LlamaForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS)
+    >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)
+
+    >>> prompt = "Hey, are you conscious? Can you talk to me?"
+    >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+    >>> # Generate
+    >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+    >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+    "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+    ```"""
+
+    output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+    output_hidden_states = (
+        output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+    )
+    return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+    # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+    outputs = self.model(
+        input_ids=input_ids,
+        modality_indicators=modality_indicators,
+        attention_mask=attention_mask,
+        position_ids=position_ids,
+        past_key_values=past_key_values,
+        inputs_embeds=inputs_embeds,
+        use_cache=use_cache,
+        output_attentions=output_attentions,
+        output_hidden_states=output_hidden_states,
+        return_dict=return_dict,
+    )
+
+    hidden_states = outputs[0]
+    if self.config.pretraining_tp > 1:
+        lm_head_slices = self.lm_head.weight.split(self.vocab_size // self.config.pretraining_tp, dim=0)
+        logits = [F.linear(hidden_states, lm_head_slices[i]) for i in range(self.config.pretraining_tp)]
+        logits = torch.cat(logits, dim=-1)
+    else:
+        logits = self.lm_head(hidden_states)
+    logits = logits.float()
+
+    loss = None
+    if labels is not None:
+        # Shift so that tokens < n predict n
+        shift_logits = logits[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+        # Flatten the tokens
+        loss_fct = CrossEntropyLoss()
+        shift_logits = shift_logits.view(-1, self.config.vocab_size)
+        shift_labels = shift_labels.view(-1)
+        # Enable model parallelism
+        shift_labels = shift_labels.to(shift_logits.device)
+        loss = loss_fct(shift_logits, shift_labels)
+
+    if not return_dict:
+        output = (logits,) + outputs[1:]
+        return (loss,) + output if loss is not None else output
+
+    return CausalLMOutputWithPast(
+        loss=loss,
+        logits=logits,
+        past_key_values=outputs.past_key_values,
+        hidden_states=outputs.hidden_states,
+        attentions=outputs.attentions,
+    )
+
+def replace_llama_modality_adaptive():
+    transformers.models.llama.configuration_llama.LlamaConfig = LlamaConfig
+    transformers.models.llama.modeling_llama.LlamaAttention = LlamaAttention
+    transformers.models.llama.modeling_llama.LlamaDecoderLayer = LlamaDecoderLayer
+    transformers.models.llama.modeling_llama.LlamaModel.forward = model_forward
+    transformers.models.llama.modeling_llama.LlamaForCausalLM.forward = causal_model_forward
+
+    
+if __name__ == "__main__":
+    replace_llama_modality_adaptive()
+    config = transformers.LlamaConfig.from_pretrained('/cpfs01/shared/public/test/vicuna-7b-v1.5/')
+    model = transformers.LlamaForCausalLM(config)
+    print(model)
\ No newline at end of file

mplug_docowl/model/modeling_mplug_docowl.py

+#    Copyright 2023 Haotian Liu & Qinghao Ye (Modified from LLaVA)
+#
+#    Licensed under the Apache License, Version 2.0 (the "License");
+#    you may not use this file except in compliance with the License.
+#    You may obtain a copy of the License at
+#
+#        http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS,
+#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#    See the License for the specific language governing permissions and
+#    limitations under the License.
+
+from abc import ABC, abstractmethod
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn as nn
+from torch.nn import CrossEntropyLoss
+
+from transformers import AutoConfig, AutoModelForCausalLM, LlamaConfig, LlamaModel, LlamaForCausalLM
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
+from .configuration_mplug_docowl import (MPLUGDocOwlConfig, MplugOwlVisionConfig, MplugDocOwlHReducerConfig)
+from .visual_encoder import MplugOwlVisionModel, MplugDocOwlHReducerModel
+from .modeling_llama2 import replace_llama_modality_adaptive
+from mplug_docowl.constants import IMAGE_TOKEN_INDEX, IGNORE_INDEX
+from icecream import ic
+
+class MPLUGDocOwlMetaModel:
+    def __init__(self, config):
+        super(MPLUGDocOwlMetaModel, self).__init__(config)
+        self.vision_model = MplugOwlVisionModel(
+            MplugOwlVisionConfig(**config.visual_config["visual_model"])
+        )
+
+        self.vision2text = MplugDocOwlHReducerModel(
+            MplugDocOwlHReducerConfig(**config.visual_config["visual_hreducer"]), config.hidden_size
+        )
+
+    def get_vision_tower(self):
+        vision_model = getattr(self, 'vision_model', None)
+        if type(vision_model) is list:
+            vision_model = vision_model[0]
+        return vision_model
+
+    def get_vision2text(self):
+        vision2text = getattr(self, 'vision2text', None)
+        if type(vision2text) is list:
+            vision2text = vision2text[0]
+        return vision2text
+
+class MPLUGDocOwlMetaForCausalLM(ABC):
+    @abstractmethod
+    def get_model(self):
+        pass
+
+    def encode_images(self, images, patch_positions):
+        image_features = self.get_model().vision_model(images).last_hidden_state
+        image_features = self.get_model().vision2text(encoder_hidden_states=image_features)
+        return image_features
+
+    def prepare_inputs_labels_for_multimodal(
+        self, input_ids, attention_mask, past_key_values, labels, images, patch_positions
+    ):
+        if images is None or input_ids.shape[1] == 1:
+            if past_key_values is not None and images is not None and input_ids.shape[1] == 1:
+                attention_mask = torch.ones((attention_mask.shape[0], past_key_values[-1][-1].shape[-2] + 1), dtype=attention_mask.dtype, device=attention_mask.device)
+            multiway_indices = torch.zeros_like(input_ids).long().to(self.device)
+            return input_ids, multiway_indices, attention_mask, past_key_values, None, labels
+        
+        if type(images) is list or images.ndim == 5:
+            concat_images = torch.cat([image for image in images], dim=0)
+            image_features = self.encode_images(concat_images, patch_positions)
+            split_sizes = [image.shape[0] for image in images]
+            image_features = torch.split(image_features, split_sizes, dim=0)
+            image_features = [x.flatten(0, 1) for x in image_features]
+        else:
+            image_features = self.encode_images(images, patch_positions) # Sum(Crop Image Number) x L x d
+
+        new_input_embeds = []
+        new_modality_indicators = []
+        new_labels = [] if labels is not None else None
+        cur_image_idx = 0
+        for batch_idx, cur_input_ids in enumerate(input_ids):
+            if (cur_input_ids == IMAGE_TOKEN_INDEX).sum() == 0:
+                # multimodal LLM, but the current sample is not multimodal
+                # FIXME: this is a hacky fix, for deepspeed zero3 to work
+                half_len = cur_input_ids.shape[0] // 2
+                cur_image_features = image_features[cur_image_idx]
+                cur_input_embeds_1 = self.get_model().embed_tokens(cur_input_ids[:half_len])
+                cur_input_embeds_2 = self.get_model().embed_tokens(cur_input_ids[half_len:])
+                cur_input_embeds = torch.cat([cur_input_embeds_1, cur_image_features[0:0], cur_input_embeds_2], dim=0)
+                new_input_embeds.append(cur_input_embeds)
+                
+                cur_modality_indicators = torch.zeros(len(cur_input_embeds)).long().to(self.device)
+                new_modality_indicators.append(cur_modality_indicators)
+                if labels is not None:
+                    new_labels.append(labels[batch_idx])
+                cur_image_idx += 1
+                continue
+            image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
+            cur_new_input_embeds = []
+            cur_modality_indicators = []
+            if labels is not None:
+                cur_labels = labels[batch_idx]
+                cur_new_labels = []
+                assert cur_labels.shape == cur_input_ids.shape
+            while image_token_indices.numel() > 0:
+                cur_image_features = image_features[cur_image_idx]
+                image_token_start = image_token_indices[0]
+                cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[:image_token_start]))
+                cur_new_input_embeds.append(cur_image_features)
+                
+                # Add modality indicator
+                assert image_token_start == len(cur_input_ids[:image_token_start])
+                cur_modality_indicators.append(torch.zeros(len(cur_input_ids[:image_token_start])).long())
+                cur_modality_indicators.append(torch.ones(len(cur_image_features)).long())
+                
+                if labels is not None:
+                    cur_new_labels.append(cur_labels[:image_token_start])
+                    cur_new_labels.append(torch.full((cur_image_features.shape[0],), IGNORE_INDEX, device=labels.device, dtype=labels.dtype))
+                    cur_labels = cur_labels[image_token_start+1:]
+                cur_image_idx += 1
+                cur_input_ids = cur_input_ids[image_token_start+1:]
+                image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
+            if cur_input_ids.numel() > 0:
+                cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids))
+                cur_modality_indicators.append(torch.zeros(len(cur_input_ids)).long())
+                if labels is not None:
+                    cur_new_labels.append(cur_labels)
+            cur_new_input_embeds = [x.to(device=self.device) for x in cur_new_input_embeds]
+            cur_new_input_embeds = torch.cat(cur_new_input_embeds, dim=0)
+            new_input_embeds.append(cur_new_input_embeds)
+            
+            # Modality
+            cur_modality_indicators = [x.to(device=self.device) for x in cur_modality_indicators]
+            cur_modality_indicators = torch.cat(cur_modality_indicators, dim=0)
+            new_modality_indicators.append(cur_modality_indicators)
+            
+            
+            if labels is not None:
+                cur_new_labels = torch.cat(cur_new_labels, dim=0)
+                new_labels.append(cur_new_labels)
+
+        if any(x.shape != new_input_embeds[0].shape for x in new_input_embeds):
+            max_len = max(x.shape[0] for x in new_input_embeds)
+            
+            # Embedding
+            new_input_embeds_align = []
+            for cur_new_embed in new_input_embeds:
+                cur_new_embed = torch.cat((cur_new_embed, torch.zeros((max_len - cur_new_embed.shape[0], cur_new_embed.shape[1]), dtype=cur_new_embed.dtype, device=cur_new_embed.device)), dim=0)
+                new_input_embeds_align.append(cur_new_embed)
+            new_input_embeds = torch.stack(new_input_embeds_align, dim=0)
+            
+            # Modality
+            new_modality_indicators_align = []
+            for cur_modality_indicator in new_modality_indicators:
+                cur_new_embed = torch.cat((cur_modality_indicator, torch.zeros(max_len - cur_modality_indicator.shape[0], dtype=cur_modality_indicator.dtype, device=cur_modality_indicator.device)), dim=0)
+                new_modality_indicators_align.append(cur_new_embed)
+            new_modality_indicators = torch.stack(new_modality_indicators_align, dim=0)
+            
+            # Label
+            if labels is not None:
+                new_labels_align = []
+                _new_labels = new_labels
+                for cur_new_label in new_labels:
+                    cur_new_label = torch.cat((cur_new_label, torch.full((max_len - cur_new_label.shape[0],), IGNORE_INDEX, dtype=cur_new_label.dtype, device=cur_new_label.device)), dim=0)
+                    new_labels_align.append(cur_new_label)
+                new_labels = torch.stack(new_labels_align, dim=0)
+            
+            # Attention Mask
+            if attention_mask is not None:
+                new_attention_mask = []
+                for cur_attention_mask, cur_new_labels, cur_new_labels_align in zip(attention_mask, _new_labels, new_labels):
+                    new_attn_mask_pad_left = torch.full((cur_new_labels.shape[0] - labels.shape[1],), True, dtype=attention_mask.dtype, device=attention_mask.device)
+                    new_attn_mask_pad_right = torch.full((cur_new_labels_align.shape[0] - cur_new_labels.shape[0],), False, dtype=attention_mask.dtype, device=attention_mask.device)
+                    cur_new_attention_mask = torch.cat((new_attn_mask_pad_left, cur_attention_mask, new_attn_mask_pad_right), dim=0)
+                    new_attention_mask.append(cur_new_attention_mask)
+                attention_mask = torch.stack(new_attention_mask, dim=0)
+                assert attention_mask.shape == new_labels.shape
+        else:
+            new_input_embeds = torch.stack(new_input_embeds, dim=0)
+            new_modality_indicators = torch.stack(new_modality_indicators, dim=0)
+            if labels is not None:
+                new_labels = torch.stack(new_labels, dim=0)
+
+            if attention_mask is not None:
+                new_attn_mask_pad_left = torch.full((attention_mask.shape[0], new_input_embeds.shape[1] - input_ids.shape[1]), True, dtype=attention_mask.dtype, device=attention_mask.device)
+                attention_mask = torch.cat((new_attn_mask_pad_left, attention_mask), dim=1)
+                assert attention_mask.shape == new_input_embeds.shape[:2]
+        return None, new_modality_indicators, attention_mask, past_key_values, new_input_embeds, new_labels
+
+
+
+class MPLUGDocOwlLlamaModel(MPLUGDocOwlMetaModel, LlamaModel):
+    config_class = MPLUGDocOwlConfig
+
+    def __init__(self, config: MPLUGDocOwlConfig):
+        super(MPLUGDocOwlLlamaModel, self).__init__(config)
+
+
+class MPLUGDocOwlLlamaForCausalLM(LlamaForCausalLM, MPLUGDocOwlMetaForCausalLM):
+    config_class = MPLUGDocOwlConfig
+
+    def __init__(self, config):
+        super(LlamaForCausalLM, self).__init__(config)
+        self.model = MPLUGDocOwlLlamaModel(config)
+
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_model(self):
+        return self.model
+
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        # modality_indicators: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        images: Optional[torch.FloatTensor] = None,
+        patch_positions: Optional[torch.LongTensor] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+
+        # print('modeling_mplug_docow2.py patch_positions:', patch_positions)
+
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+        input_ids, modality_indicators, attention_mask, past_key_values, inputs_embeds, labels = \
+            self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images, patch_positions)
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model(
+            input_ids=input_ids,
+            modality_indicators=modality_indicators,
+            attention_mask=attention_mask,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model/pipeline parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    def prepare_inputs_for_generation(
+        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
+    ):
+        if past_key_values:
+            input_ids = input_ids[:, -1:]
+
+        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
+        if inputs_embeds is not None and past_key_values is None:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        model_inputs.update(
+            {
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "attention_mask": attention_mask,
+                "images": kwargs.get("images", None),
+                "patch_positions": kwargs.get("patch_positions", None),
+            }
+        )
+        return model_inputs
+
+AutoConfig.register("mplug_docowl", MPLUGDocOwlConfig)
+AutoModelForCausalLM.register(MPLUGDocOwlConfig, MPLUGDocOwlLlamaForCausalLM)
+
+replace_llama_modality_adaptive()
+    
\ No newline at end of file

mplug_docowl/model/utils.py

+from transformers import AutoConfig
+
+
+def auto_upgrade(config):
+    cfg = AutoConfig.from_pretrained(config)
+    if 'mplug_owl2' in config and 'mplug_owl2' not in cfg.model_type:
+        assert cfg.model_type == 'mplug_owl2'
+        print("You are using newer LLaVA code base, while the checkpoint of v0 is from older code base.")
+        print("You must upgrade the checkpoint to the new code base (this can be done automatically).")
+        confirm = input("Please confirm that you want to upgrade the checkpoint. [Y/N]")
+        if confirm.lower() in ["y", "yes"]:
+            print("Upgrading checkpoint...")
+            assert len(cfg.architectures) == 1
+            setattr(cfg.__class__, "model_type", "mplug_owl2")
+            cfg.architectures[0] = 'LlavaLlamaForCausalLM'
+            cfg.save_pretrained(config)
+            print("Checkpoint upgraded.")
+        else:
+            print("Checkpoint upgrade aborted.")
+            exit(1)
\ No newline at end of file

mplug_docowl/model/visual_encoder.py

+import math
+from typing import Any, Optional, Tuple, Union
+
+from transformers.modeling_outputs import BaseModelOutput, BaseModelOutputWithPooling, BaseModelOutputWithPastAndCrossAttentions
+from transformers.modeling_utils import PreTrainedModel
+from transformers.pytorch_utils import find_pruneable_heads_and_indices, prune_linear_layer
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.utils.checkpoint
+from icecream import ic
+import einops
+from einops import rearrange
+
+def get_abs_pos(abs_pos, tgt_size):
+    # abs_pos: L, C
+    # tgt_size: M
+    # return: M, C
+    src_size = int(math.sqrt(abs_pos.size(0)))
+    tgt_size = int(math.sqrt(tgt_size))
+    dtype = abs_pos.dtype
+
+    if src_size != tgt_size:
+        return F.interpolate(
+            abs_pos.float().reshape(1, src_size, src_size, -1).permute(0, 3, 1, 2),
+            size=(tgt_size, tgt_size),
+            mode="bicubic",
+            align_corners=False,
+        ).permute(0, 2, 3, 1).flatten(0, 2).to(dtype=dtype)
+    else:
+        return abs_pos
+
+# https://github.com/facebookresearch/mae/blob/efb2a8062c206524e35e47d04501ed4f544c0ae8/util/pos_embed.py#L20
+def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False):
+    """
+    grid_size: int of the grid height and width
+    return:
+    pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/ or w/o cls_token)
+    """
+    grid_h = np.arange(grid_size, dtype=np.float32)
+    grid_w = np.arange(grid_size, dtype=np.float32)
+    grid = np.meshgrid(grid_w, grid_h)  # here w goes first
+    grid = np.stack(grid, axis=0)
+
+    grid = grid.reshape([2, 1, grid_size, grid_size])
+    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
+    if cls_token:
+        pos_embed = np.concatenate([np.zeros([1, embed_dim]), pos_embed], axis=0)
+    return pos_embed
+
+
+def get_2d_sincos_pos_embed_from_grid(embed_dim, grid):
+    assert embed_dim % 2 == 0
+
+    # use half of dimensions to encode grid_h
+    emb_h = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0])  # (H*W, D/2)
+    emb_w = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1])  # (H*W, D/2)
+
+    emb = np.concatenate([emb_h, emb_w], axis=1) # (H*W, D)
+    return emb
+
+
+def get_1d_sincos_pos_embed_from_grid(embed_dim, pos):
+    """
+    embed_dim: output dimension for each position
+    pos: a list of positions to be encoded: size (M,)
+    out: (M, D)
+    """
+    assert embed_dim % 2 == 0
+    omega = np.arange(embed_dim // 2, dtype=np.float32)
+    omega /= embed_dim / 2.
+    omega = 1. / 10000**omega  # (D/2,)
+
+    pos = pos.reshape(-1)  # (M,)
+    out = np.einsum('m,d->md', pos, omega)  # (M, D/2), outer product
+
+    emb_sin = np.sin(out) # (M, D/2)
+    emb_cos = np.cos(out) # (M, D/2)
+
+    emb = np.concatenate([emb_sin, emb_cos], axis=1)  # (M, D)
+    return emb
+
+
+
+class MplugOwlVisionEmbeddings(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.image_size = config.image_size
+        self.patch_size = config.patch_size
+
+        self.cls_token = nn.Parameter(torch.randn(1, 1, self.hidden_size))
+
+        self.patch_embed = nn.Conv2d(
+            in_channels=3,
+            out_channels=self.hidden_size,
+            kernel_size=self.patch_size,
+            stride=self.patch_size,
+            bias=False,
+        )
+
+        self.num_patches = (self.image_size // self.patch_size) ** 2
+
+        self.position_embedding = nn.Parameter(torch.randn(1, self.num_patches + 1, self.hidden_size))
+
+        self.pre_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+    def forward(self, pixel_values: torch.FloatTensor) -> torch.Tensor:
+        batch_size = pixel_values.size(0)
+        image_embeds = self.patch_embed(pixel_values)
+        image_embeds = image_embeds.flatten(2).transpose(1, 2)
+
+        class_embeds = self.cls_token.expand(batch_size, 1, -1).to(image_embeds.dtype)
+        embeddings = torch.cat([class_embeds, image_embeds], dim=1)
+        embeddings = embeddings + self.position_embedding[:, : embeddings.size(1)].to(image_embeds.dtype)
+        embeddings = self.pre_layernorm(embeddings)
+        return embeddings
+
+
+
+class MplugOwlVisionAttention(nn.Module):
+    """Multi-headed attention from 'Attention Is All You Need' paper"""
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        if self.head_dim * self.num_heads != self.hidden_size:
+            raise ValueError(
+                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size} and `num_heads`:"
+                f" {self.num_heads})."
+            )
+        self.scale = self.head_dim**-0.5
+        self.dropout = nn.Dropout(config.attention_dropout)
+
+        self.query_key_value = nn.Linear(self.hidden_size, 3 * self.hidden_size)
+        self.dense = nn.Linear(self.hidden_size, self.hidden_size)
+
+    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
+        return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        head_mask: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        """Input shape: Batch x Time x Channel"""
+
+        bsz, seq_len, embed_dim = hidden_states.size()
+
+        mixed_qkv = self.query_key_value(hidden_states)
+
+        mixed_qkv = mixed_qkv.reshape(bsz, seq_len, self.num_heads, 3, embed_dim // self.num_heads).permute(
+            3, 0, 2, 1, 4
+        )  # [3, b, np, sq, hn]
+        query_states, key_states, value_states = (
+            mixed_qkv[0],
+            mixed_qkv[1],
+            mixed_qkv[2],
+        )
+        # if self.config.use_flash_attn and flash_attn_func is not None:
+        if False:
+            # [b*sq, np, hn]
+            query_states = query_states.permute(0, 2, 1, 3).contiguous()
+            query_states = query_states.view(query_states.size(0) * query_states.size(1), query_states.size(2), -1)
+
+            key_states = key_states.permute(0, 2, 1, 3).contiguous()
+            key_states = key_states.view(key_states.size(0) * key_states.size(1), key_states.size(2), -1)
+
+            value_states = value_states.permute(0, 2, 1, 3).contiguous()
+            value_states = value_states.view(value_states.size(0) * value_states.size(1), value_states.size(2), -1)
+
+            cu_seqlens = torch.arange(
+                0, (bsz + 1) * seq_len, step=seq_len, dtype=torch.int32, device=query_states.device
+            )
+
+            context_layer = flash_attn_func(
+                query_states,
+                key_states,
+                value_states,
+                cu_seqlens,
+                cu_seqlens,
+                seq_len,
+                seq_len,
+                self.dropout if self.training else 0.0,
+                softmax_scale=self.scale,
+                causal=False,
+                return_attn_probs=False,
+            )
+            # [b*sq, np, hn] => [b, sq, np, hn]
+            context_layer = context_layer.view(bsz, seq_len, context_layer.size(1), context_layer.size(2))
+        else:
+            # Take the dot product between "query" and "key" to get the raw attention scores.
+            attention_scores = torch.matmul(query_states, key_states.transpose(-1, -2))
+
+            attention_scores = attention_scores * self.scale
+
+            # Normalize the attention scores to probabilities.
+            attention_probs = torch.softmax(attention_scores, dim=-1)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.dropout(attention_probs)
+
+            # Mask heads if we want to
+            if head_mask is not None:
+                attention_probs = attention_probs * head_mask
+
+            context_layer = torch.matmul(attention_probs, value_states).permute(0, 2, 1, 3)
+
+        new_context_layer_shape = context_layer.size()[:-2] + (self.hidden_size,)
+        context_layer = context_layer.reshape(new_context_layer_shape)
+
+        output = self.dense(context_layer)
+
+        outputs = (output, attention_probs) if output_attentions else (output, None)
+
+        return outputs
+
+
+class QuickGELU(nn.Module):
+    def forward(self, x: torch.Tensor):
+        return x * torch.sigmoid(1.702 * x)
+
+
+class MplugOwlMLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.activation_fn = QuickGELU()
+        self.fc1 = nn.Linear(config.hidden_size, config.intermediate_size)
+        self.fc2 = nn.Linear(config.intermediate_size, config.hidden_size)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.fc1(hidden_states)
+        hidden_states = self.activation_fn(hidden_states)
+        hidden_states = self.fc2(hidden_states)
+        return hidden_states
+
+
+class MplugOwlVisionEncoderLayer(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.hidden_size = config.hidden_size
+        self.self_attn = MplugOwlVisionAttention(config)
+        self.input_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+        self.mlp = MplugOwlMLP(config)
+        self.post_attention_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: torch.Tensor,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.FloatTensor]:
+        """
+        Args:
+            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
+            attention_mask (`torch.FloatTensor`): attention mask of size
+                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
+                `(config.encoder_attention_heads,)`.
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+        """
+        residual = hidden_states
+
+        hidden_states = self.input_layernorm(hidden_states)
+        hidden_states, attn_weights = self.self_attn(
+            hidden_states=hidden_states,
+            head_mask=attention_mask,
+            output_attentions=output_attentions,
+        )
+        hidden_states = hidden_states + residual
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+
+        hidden_states = hidden_states + residual
+
+        outputs = (hidden_states,)
+
+        if output_attentions:
+            outputs += (attn_weights,)
+
+        return outputs
+    
+    
+class MplugOwlVisionEncoder(nn.Module):
+    """
+    Transformer encoder consisting of `config.num_hidden_layers` self attention layers. Each layer is a
+    [`MplugOwlVisionEncoderLayer`].
+
+    Args:
+        config (`MplugOwlVisionConfig`):
+            The corresponding vision configuration for the `MplugOwlEncoder`.
+    """
+
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.layers = nn.ModuleList([MplugOwlVisionEncoderLayer(config) for _ in range(config.num_hidden_layers)])
+        self.gradient_checkpointing = True
+
+    def forward(
+        self,
+        inputs_embeds,
+        attention_mask: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutput]:
+        r"""
+        Args:
+            inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
+                Embedded representation of the inputs. Should be float, not int tokens.
+            attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
+                - 1 for tokens that are **not masked**,
+                - 0 for tokens that are **masked**.
+
+                [What are attention masks?](../glossary#attention-mask)
+            output_attentions (`bool`, *optional*):
+                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
+                returned tensors for more detail.
+            output_hidden_states (`bool`, *optional*):
+                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
+                for more detail.
+            return_dict (`bool`, *optional*):
+                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
+        """
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        encoder_states = () if output_hidden_states else None
+        all_attentions = () if output_attentions else None
+
+        hidden_states = inputs_embeds
+        for idx, encoder_layer in enumerate(self.layers):
+            if output_hidden_states:
+                encoder_states = encoder_states + (hidden_states,)
+            if self.gradient_checkpointing and self.training:
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        return module(*inputs, output_attentions)
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(encoder_layer),
+                    hidden_states,
+                    attention_mask,
+                )
+            else:
+                layer_outputs = encoder_layer(
+                    hidden_states,
+                    attention_mask,
+                    output_attentions=output_attentions,
+                )
+
+            hidden_states = layer_outputs[0]
+
+            if output_attentions:
+                all_attentions = all_attentions + (layer_outputs[1],)
+
+        if output_hidden_states:
+            encoder_states = encoder_states + (hidden_states,)
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None)
+        return BaseModelOutput(
+            last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions
+        )
+
+
+class MplugOwlVisionModel(PreTrainedModel):
+    main_input_name = "pixel_values"
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.config = config
+        self.hidden_size = config.hidden_size
+
+        self.embeddings = MplugOwlVisionEmbeddings(config)
+        self.encoder = MplugOwlVisionEncoder(config)
+        self.post_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps)
+
+        self.post_init()
+
+
+    def forward(
+        self,
+        pixel_values: Optional[torch.FloatTensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPooling]:
+        r"""
+        Returns:
+
+        """
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if pixel_values is None:
+            raise ValueError("You have to specify pixel_values")
+
+        hidden_states = self.embeddings(pixel_values)
+
+        encoder_outputs = self.encoder(
+            inputs_embeds=hidden_states,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        last_hidden_state = encoder_outputs[0]
+        last_hidden_state = self.post_layernorm(last_hidden_state)
+
+        pooled_output = last_hidden_state[:, 0, :]
+        pooled_output = self.post_layernorm(pooled_output)
+
+        if not return_dict:
+            return (last_hidden_state, pooled_output) + encoder_outputs[1:]
+
+        return BaseModelOutputWithPooling(
+            last_hidden_state=last_hidden_state,
+            pooler_output=pooled_output,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+        )
+
+    def get_input_embeddings(self):
+        return self.embeddings
+
+
+class MplugDocOwlHReducerModel(PreTrainedModel):
+    def __init__(self, config, language_hidden_size):
+        super().__init__(config)
+        self.config = config
+        self.ln_q = torch.nn.LayerNorm(self.config.hidden_size, eps=1e-6)
+        self.conv_shape = (int(self.config.conv_shape.split('x')[0]), int(self.config.conv_shape.split('x')[1])) # 
+        self.conv_patch=self.conv_shape[0]*self.conv_shape[1]
+        ## feature interaction with a conv layer
+        self.reducer_before = torch.nn.Sequential(
+            nn.Conv2d(self.config.hidden_size, self.conv_patch*self.config.hidden_size, kernel_size=self.conv_shape, stride=self.conv_shape, bias=True),
+            nn.GELU()
+        )
+        ## reduce visual feature length with a conv layer
+        self.reducer = nn.Conv2d(self.config.hidden_size, self.config.hidden_size, kernel_size=self.conv_shape, stride=self.conv_shape, bias=True)    
+        ## align visual features with language embedding with fc
+        self.visual_fc = torch.nn.Linear(self.config.hidden_size, language_hidden_size)
+        self.vit_eos = torch.nn.Parameter(torch.randn(1, 1, language_hidden_size))
+
+        self.post_init()
+
+    def forward(
+        self,
+        encoder_hidden_states=None
+    ):
+        r"""
+        encoder_hidden_states  (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, `optional`):
+            batch_size is the number of all images (global+crop) in a batch
+            Sequence of hidden-states at the output of the last layer of the encoder.
+        """
+        encoder_hidden_states = encoder_hidden_states[:,1:,:] # remove the first cls token 
+        B, L, C = encoder_hidden_states.shape # B, 1024=(448/14)^2, 1024
+
+        ## feature interaction with a conv layer
+        encoder_hidden_states = rearrange(encoder_hidden_states, 'B (H W) D -> B D H W', H=int(math.sqrt(L)))
+        hidden_states = self.reducer_before(encoder_hidden_states) # B 4D H W/4
+        ## reduce seq length with a conv layer
+        hidden_states = rearrange(hidden_states, 'B (X D) H W -> B D H (W X)', X=self.conv_patch) # B 4D H W/4 -> B D H W
+        sequence_output = self.reducer(hidden_states) # B,C,H,W -> B,C,H/conv_shape[0],W/(conv_shape[1])
+        sequence_output = sequence_output.flatten(2).transpose(1, 2)  # B,C,H/conv_shape[0],W/(conv_shape[1]) -> B,C,L/conv_patch -> B,L/conv_patch,C
+        sequence_output = sequence_output.transpose(0, 1).contiguous() # L/conv_patch, B, C
+        ## align visual features with language embedding with fc
+        sequence_output = self.visual_fc(sequence_output) # L/conv_patch, B, h
+        sequence_output = sequence_output.transpose(0, 1).contiguous() # B, s/4, h
+        sequence_output = torch.cat([sequence_output, self.vit_eos.repeat(B, 1, 1)], dim=1)
+
+        return sequence_output
+

mplug_docowl/processor.py

+from einops import rearrange, repeat
+import torch
+from torchvision import transforms
+from PIL import Image, ImageFile
+import random
+from torchvision.ops.boxes import box_area
+
+from torchvision.transforms.transforms import InterpolationMode
+from torchvision.transforms import functional as F
+import numpy as np
+from icecream import ic
+
+ImageFile.LOAD_TRUNCATED_IMAGES = True
+ImageFile.MAX_IMAGE_PIXELS = None
+Image.MAX_IMAGE_PIXELS = None
+
+def box_iou(boxes1, area1, boxes2, eps=1e-5):
+    area2 = box_area(boxes2)
+
+    lt = torch.max(boxes1[:, None, :2], boxes2[:, :2])  # [N,M,2]
+    rb = torch.min(boxes1[:, None, 2:], boxes2[:, 2:])  # [N,M,2]
+
+    wh = (rb - lt).clamp(min=0)  # [N,M,2]
+    inter = wh[:, :, 0] * wh[:, :, 1]  # [N,M]
+
+    union = area1[:, None] + area2 - inter
+
+    iou = inter / (union+eps)
+    return iou, union
+
+def anchor_rank(anchors, anchors_areas, input_image_size, eps=1e-5):
+    # anchors x1 y1 x2 y2
+
+    # image_size: (h, w)
+    # xyxy
+    input_image_bbox = torch.tensor([0, 0, input_image_size[1], input_image_size[0]]).unsqueeze(0)
+
+    boxes1 = anchors
+    boxes2 = input_image_bbox
+    boxes3 = anchors.clone()
+    # y2
+    boxes3[:,3] = input_image_size[0]/input_image_size[1]*anchors[:,2] # 用于算分辨率无关的iou
+    
+    area1 = anchors_areas
+    
+    iou, _ = box_iou(boxes1, area1, boxes2)
+    iou = iou.squeeze(1)
+    shape_iou, _ = box_iou(boxes1, area1, boxes3)
+    shape_iou = shape_iou.diag()
+    # 优先匹配形状接近 再匹配分辨率接近
+    index = torch.argmax(shape_iou*100+iou,dim=0)
+    return index
+
+class AnchorResize(torch.nn.Module):
+
+    def __init__(self, image_size, anchors, interpolation=InterpolationMode.BILINEAR, antialias=None):
+        super().__init__()
+        # xyxy
+        self.anchors = torch.tensor(
+            [[0, 0, _[1]*image_size[1], _[0]*image_size[0]] 
+            for _ in anchors], requires_grad=False
+        )
+        
+        self.anchor_areas = box_area(self.anchors)
+
+        self.interpolation = interpolation
+        self.antialias = antialias
+
+    def forward(self, img, skip_resize=False):
+        """
+        Args:
+            img (PIL Image or Tensor): Image to be scaled.
+
+        Returns:
+            PIL Image or Tensor: Rescaled image.
+        """
+        selected_anchor = anchor_rank(self.anchors, self.anchor_areas, (img.size[1], img.size[0]))
+        target_size = self.anchors[selected_anchor][2:].tolist() # w,h
+        if skip_resize:
+            # for debug
+            return selected_anchor
+        return F.resize(img, [target_size[1],target_size[0]], self.interpolation, max_size=None, antialias=self.antialias), selected_anchor
+
+    def __repr__(self) -> str:
+        detail = f"(size={self.image_size}, anchor={self.anchors}, interpolation={self.interpolation.value}, antialias={self.antialias})"
+        return f"{self.__class__.__name__}{detail}"
+
+grid_dict = {
+    'grid_1':[
+        (1,1)],
+    'grid_4':[
+        (1,1),
+        (1,2),(2,1),
+        (1,3),(3,1),
+        (2,2),(1,4),(4,1)],
+    'grid_9':[
+        (1,1),
+        (1,2),(2,1),
+        (1,3),(3,1),
+        (2,2),(1,4),(4,1),
+        (1,5),(5,1),
+        (1,6),(6,1),(2,3),(3,2),
+        (1,7),(7,1),
+        (4,2),(2,4),(1,8),(8,1),
+        (3,3),(1,9),(9,1)],
+    'grid_3x3':[
+        (3,3)],
+    'grid_20':[
+        (1, 1), 
+        (1, 2), (2, 1), 
+        (1, 3), (3, 1), (1, 4), (2, 2), (4, 1), 
+        (1, 5), (5, 1), 
+        (1, 6), (2, 3), (3, 2), (6, 1), 
+        (1, 7), (7, 1), 
+        (1, 8), (2, 4), (4, 2), (8, 1), 
+        (1, 9), (3, 3), (9, 1), 
+        (1, 10), (2, 5), (5, 2), (10, 1), 
+        (1, 11), (11, 1), 
+        (2, 6), (3, 4), (4, 3), (6, 2), 
+        (2, 7), (7, 2), 
+        (3, 5), (5, 3), 
+        (2, 8), (4, 4), (8, 2), 
+        (2, 9), (3, 6), (6, 3), (9, 2), 
+        (2, 10), (4, 5), (5, 4), (10, 2)]
+}
+
+class DocProcessor():
+    def __init__(self, image_size=224, anchors='grid_9', add_global_img=True, add_textual_crop_indicator=False):
+        self.add_global_img = add_global_img
+        self.add_textual_crop_indicator = add_textual_crop_indicator
+        self.media_token= "<|image|>"
+        # h,w
+        if isinstance(image_size, int):
+            image_size = (image_size, image_size)
+        self.image_size = image_size
+        # h,w
+        anchors = grid_dict[anchors]
+        self.anchors = [tuple(_) for _ in anchors]
+        self.anchor_max = max([max(_) for _ in self.anchors])
+        # xywh -> xyxy
+        self.resizer = AnchorResize(image_size=image_size, anchors=anchors, interpolation=InterpolationMode.BICUBIC)
+        self.old_resizer = transforms.Resize(image_size,interpolation=InterpolationMode.BICUBIC)
+        self.image_transform = transforms.Compose([
+            transforms.ToTensor(),
+            transforms.Normalize((0.48145466, 0.4578275, 0.40821073), (0.26862954, 0.26130258, 0.27577711)),
+        ])
+    
+    def _process_image(self, images):
+        new_images = []
+        new_patch_position = []
+        num_image_mult = []
+        for image in images:
+            if self.add_global_img:
+                nocut_image = self.image_transform(self.old_resizer(image)).unsqueeze(0)
+                
+            image, selected_anchor = self.resizer(image)
+            image_input = self.image_transform(image) # h,w,3 -> 3,h,w
+            # rearrange(x,'B C (n1 h) (n2 w) -> (B n1 n2) C h w', n1=self.down_sample[0], n2=self.down_sample[1])
+            image_input = rearrange(image_input, 'C (num_h h) (num_w w) -> (num_h num_w) C h w', h=self.image_size[0], w=self.image_size[1])
+
+            if self.add_global_img:
+                image_input = torch.cat([nocut_image, image_input], dim=0)
+
+            anchor = self.anchors[selected_anchor] # w,h
+            patch_position = torch.cat([
+                repeat(torch.arange(anchor[0]), 'num_h -> num_h num_w 1', num_w=anchor[1]),
+                repeat(torch.arange(anchor[1]), 'num_w -> num_h num_w 1', num_h=anchor[0])],dim=2)
+            patch_position = rearrange(patch_position, 'num_h num_w p-> (num_h num_w) p', p=2) # num_patch, (ph,pw)
+
+            if self.add_global_img:
+                patch_position = torch.cat([torch.ones(1,2).long()*self.anchor_max, patch_position], dim=0)
+
+            new_images.append(image_input)
+            new_patch_position.append(patch_position)
+            num_image_mult.append(patch_position.shape[0])
+
+        new_images = torch.cat(new_images,dim=0)
+        new_patch_position = torch.cat(new_patch_position, dim=0)
+        return new_images, new_patch_position, num_image_mult
+
+    def __call__(self, images=None, query=None):
+        assert images is not None
+
+        if not isinstance(images, list):
+            images = [images]
+        image_pils = []
+        for image in images:
+            if isinstance(image, str):
+                image = Image.open(image).convert('RGB')
+            else:
+                image = image.convert('RGB')
+            # ic(image.size)
+            image_pils.append(image)
+
+        image_data, patch_position, num_image_mult = self._process_image(image_pils)
+
+        assert self.media_token in query
+        text_list = query.split(self.media_token)
+        text = text_list[0]
+        image_token_ptr = 0
+        for next_text in text_list[1:]:
+            if self.add_textual_crop_indicator:
+                # generate image placeholders with interleaved texutual crop indicator
+                # e.g. <global_img><|image|><crop_img_row0_col0><|image|><crop_img_row0_col1><|image|>...
+                for patch_pos in patch_position.tolist():
+                    # global non-crop image
+                    if patch_pos[0] == self.anchor_max and patch_pos[1] == self.anchor_max:
+                        text += '<global_img><|image|>'
+                    else:
+                        row_col = 'row'+str(patch_pos[0])+'_col'+str(patch_pos[1])
+                        text += '<crop_img_'+row_col+'><|image|>'
+            else: 
+                # generate successive image placeholders for a image, 1 crop img == 1 <|image|>
+                text += '<|image|>'*num_image_mult[image_token_ptr]
+            text += next_text
+            image_token_ptr += 1
+
+        return image_data, patch_position, text
\ No newline at end of file

mplug_docowl/train/llama_flash_attn_monkey_patch.py

+from typing import Optional, Tuple
+import warnings
+
+import torch
+
+import transformers
+from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, repeat_kv
+
+try:
+    from flash_attn.flash_attn_interface import flash_attn_unpadded_qkvpacked_func
+except ImportError:
+    from flash_attn.flash_attn_interface import flash_attn_varlen_qkvpacked_func as flash_attn_unpadded_qkvpacked_func
+from flash_attn.bert_padding import unpad_input, pad_input
+
+
+def forward(
+    self,
+    hidden_states: torch.Tensor,
+    modality_indicators: torch.Tensor,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.Tensor] = None,
+    past_key_value: Optional[Tuple[torch.Tensor]] = None,
+    output_attentions: bool = False,
+    use_cache: bool = False,
+    padding_mask: bool = None,
+) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    if output_attentions:
+        warnings.warn(
+            "Output attentions is not supported for patched `LlamaAttention`, returning `None` instead."
+        )
+
+    bsz, q_len, _ = hidden_states.size()
+
+    query_states = (
+        self.q_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    key_states = (
+        self.k_proj(hidden_states, modality_indicators)
+        .view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    value_states = (
+        self.v_proj(hidden_states, modality_indicators)
+        .view(bsz, q_len, self.num_key_value_heads, self.head_dim)
+        .transpose(1, 2)
+    )  # shape: (b, num_heads, s, head_dim)
+
+    kv_seq_len = key_states.shape[-2]
+    if past_key_value is not None:
+        kv_seq_len += past_key_value[0].shape[-2]
+
+    cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+    query_states, key_states = apply_rotary_pos_emb(
+        query_states, key_states, cos, sin, position_ids
+    )
+
+    if past_key_value is not None:
+        # reuse k, v
+        key_states = torch.cat([past_key_value[0], key_states], dim=2)
+        value_states = torch.cat([past_key_value[1], value_states], dim=2)
+
+    past_key_value = (key_states, value_states) if use_cache else None
+
+    # repeat k/v heads if n_kv_heads < n_heads
+    key_states = repeat_kv(key_states, self.num_key_value_groups)
+    value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+    # Transform the data into the format required by flash attention
+    qkv = torch.stack([query_states, key_states, value_states], dim=2)
+    qkv = qkv.transpose(1, 3)  # shape: [b, s, 3, num_heads, head_dim]
+    key_padding_mask = attention_mask
+
+    if key_padding_mask is None:
+        qkv = qkv.reshape(-1, 3, self.num_heads, self.head_dim)
+        cu_q_lens = torch.arange(
+            0, (bsz + 1) * q_len, step=q_len, dtype=torch.int32, device=qkv.device
+        )
+        max_s = q_len
+        output = flash_attn_unpadded_qkvpacked_func(
+            qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True
+        )
+        output = output.view(bsz, q_len, -1)
+    else:
+        qkv = qkv.reshape(bsz, q_len, -1)
+        qkv, indices, cu_q_lens, max_s = unpad_input(qkv, key_padding_mask)
+        qkv = qkv.view(-1, 3, self.num_heads, self.head_dim)
+        output_unpad = flash_attn_unpadded_qkvpacked_func(
+            qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True
+        )
+        output_unpad = output_unpad.reshape(-1, self.num_heads * self.head_dim)
+        output = pad_input(output_unpad, indices, bsz, q_len)
+
+    return self.o_proj(output), None, past_key_value
+
+
+# Disable the transformation of the attention mask in LlamaModel as the flash attention
+# requires the attention mask to be the same as the key_padding_mask
+def _prepare_decoder_attention_mask(
+    self, attention_mask, input_shape, inputs_embeds, past_key_values_length
+):
+    # [bsz, seq_len]
+    return attention_mask
+
+
+def replace_llama_attn_with_flash_attn():
+    cuda_major, cuda_minor = torch.cuda.get_device_capability()
+    if cuda_major < 8:
+        warnings.warn(
+            "Flash attention is only supported on A100 or H100 GPU during training due to head dim > 64 backward."
+            "ref: https://github.com/HazyResearch/flash-attention/issues/190#issuecomment-1523359593"
+        )
+    transformers.models.llama.modeling_llama.LlamaModel._prepare_decoder_attention_mask = (
+        _prepare_decoder_attention_mask
+    )
+    transformers.models.llama.modeling_llama.LlamaAttention.forward = forward
\ No newline at end of file

mplug_docowl/train/mplug_docowl_trainer.py

+import os
+import torch
+
+from torch.utils.data import Sampler
+
+from transformers import Trainer
+from transformers.trainer import (
+    is_sagemaker_mp_enabled,
+    get_parameter_names,
+    has_length,
+    ALL_LAYERNORM_LAYERS,
+    ShardedDDPOption,
+    logger,
+)
+from typing import List, Optional
+from icecream import ic
+
+def maybe_zero_3(param, ignore_status=False, name=None):
+    from deepspeed import zero
+    from deepspeed.runtime.zero.partition_parameters import ZeroParamStatus
+    if hasattr(param, "ds_id"):
+        if param.ds_status == ZeroParamStatus.NOT_AVAILABLE:
+            if not ignore_status:
+                print(name, 'no ignore status')
+        with zero.GatheredParameters([param]):
+            param = param.data.detach().cpu().clone()
+    else:
+        param = param.detach().cpu().clone()
+    return param
+
+
+def get_mm_adapter_state_maybe_zero_3(named_params, keys_to_match):
+    to_return = {k: t for k, t in named_params if any(key_match in k for key_match in keys_to_match)}
+    to_return = {k: maybe_zero_3(v, ignore_status=True, name=k).cpu() for k, v in to_return.items()}
+    return to_return
+
+
+def split_to_even_chunks(indices, lengths, num_chunks):
+    """
+    Split a list of indices into `chunks` chunks of roughly equal lengths.
+    """
+
+    if len(indices) % num_chunks != 0:
+        return [indices[i::num_chunks] for i in range(num_chunks)]
+
+    num_indices_per_chunk = len(indices) // num_chunks
+
+    chunks = [[] for _ in range(num_chunks)]
+    chunks_lengths = [0 for _ in range(num_chunks)]
+    for index in indices:
+        shortest_chunk = chunks_lengths.index(min(chunks_lengths))
+        chunks[shortest_chunk].append(index)
+        chunks_lengths[shortest_chunk] += lengths[index]
+        if len(chunks[shortest_chunk]) == num_indices_per_chunk:
+            chunks_lengths[shortest_chunk] = float("inf")
+
+    return chunks
+
+
+def get_modality_length_grouped_indices(lengths, batch_size, world_size, generator=None):
+    # We need to use torch for the random part as a distributed sampler will set the random seed for torch.
+    assert all(l != 0 for l in lengths), "Should not have zero length."
+    if all(l > 0 for l in lengths) or all(l < 0 for l in lengths):
+        # all samples are in the same modality
+        return get_length_grouped_indices(lengths, batch_size, world_size, generator=generator)
+    mm_indices, mm_lengths = zip(*[(i, l) for i, l in enumerate(lengths) if l > 0])
+    lang_indices, lang_lengths = zip(*[(i, -l) for i, l in enumerate(lengths) if l < 0])
+
+    mm_shuffle = [mm_indices[i] for i in get_length_grouped_indices(mm_lengths, batch_size, world_size, generator=None)]
+    lang_shuffle = [lang_indices[i] for i in get_length_grouped_indices(lang_lengths, batch_size, world_size, generator=None)]
+    megabatch_size = world_size * batch_size
+    mm_megabatches = [mm_shuffle[i : i + megabatch_size] for i in range(0, len(mm_shuffle), megabatch_size)]
+    lang_megabatches = [lang_shuffle[i : i + megabatch_size] for i in range(0, len(lang_shuffle), megabatch_size)]
+
+    last_mm = mm_megabatches[-1]
+    last_lang = lang_megabatches[-1]
+    additional_batch = last_mm + last_lang
+    megabatches = mm_megabatches[:-1] + lang_megabatches[:-1]
+    megabatch_indices = torch.randperm(len(megabatches), generator=generator)
+    megabatches = [megabatches[i] for i in megabatch_indices]
+
+    if len(additional_batch) > 0:
+        megabatches.append(sorted(additional_batch))
+
+    return [i for megabatch in megabatches for i in megabatch]
+
+
+def get_length_grouped_indices(lengths, batch_size, world_size, generator=None, merge=True):
+    # We need to use torch for the random part as a distributed sampler will set the random seed for torch.
+    indices = torch.randperm(len(lengths), generator=generator)
+    megabatch_size = world_size * batch_size
+    megabatches = [indices[i : i + megabatch_size].tolist() for i in range(0, len(lengths), megabatch_size)]
+    megabatches = [sorted(megabatch, key=lambda i: lengths[i], reverse=True) for megabatch in megabatches]
+    megabatches = [split_to_even_chunks(megabatch, lengths, world_size) for megabatch in megabatches]
+
+    return [i for megabatch in megabatches for batch in megabatch for i in batch]
+
+
+class LengthGroupedSampler(Sampler):
+    r"""
+    Sampler that samples indices in a way that groups together features of the dataset of roughly the same length while
+    keeping a bit of randomness.
+    """
+
+    def __init__(
+        self,
+        batch_size: int,
+        world_size: int,
+        lengths: Optional[List[int]] = None,
+        generator=None,
+        group_by_modality: bool = False,
+    ):
+        if lengths is None:
+            raise ValueError("Lengths must be provided.")
+
+        self.batch_size = batch_size
+        self.world_size = world_size
+        self.lengths = lengths
+        self.generator = generator
+        self.group_by_modality = group_by_modality
+
+    def __len__(self):
+        return len(self.lengths)
+
+    def __iter__(self):
+        if self.group_by_modality:
+            indices = get_modality_length_grouped_indices(self.lengths, self.batch_size, self.world_size, generator=self.generator)
+        else:
+            indices = get_length_grouped_indices(self.lengths, self.batch_size, self.world_size, generator=self.generator)
+        return iter(indices)
+
+
+class MPLUGDocOwlTrainer(Trainer):
+
+    def _get_train_sampler(self) -> Optional[torch.utils.data.Sampler]:
+        if self.train_dataset is None or not has_length(self.train_dataset):
+            return None
+
+        if self.args.group_by_modality_length:
+            lengths = self.train_dataset.modality_lengths
+            return LengthGroupedSampler(
+                self.args.train_batch_size,
+                world_size=self.args.world_size * self.args.gradient_accumulation_steps,
+                lengths=lengths,
+                group_by_modality=True,
+            )
+        else:
+            return super()._get_train_sampler()
+
+    def create_optimizer(self):
+        """
+        Setup the optimizer.
+
+        We provide a reasonable default that works well. If you want to use something else, you can pass a tuple in the
+        Trainer's init through `optimizers`, or subclass and override this method in a subclass.
+        """
+        if is_sagemaker_mp_enabled():
+            return super().create_optimizer()
+        if self.sharded_ddp == ShardedDDPOption.SIMPLE:
+            return super().create_optimizer()
+
+        opt_model = self.model
+
+        if self.optimizer is None:
+            decay_parameters = get_parameter_names(opt_model, ALL_LAYERNORM_LAYERS)
+            decay_parameters = [name for name in decay_parameters if "bias" not in name]
+            if self.args.vision2text_lr is not None:
+                projector_parameters = [name for name, _ in opt_model.named_parameters() if "vision2text_lr" in name]
+                optimizer_grouped_parameters = [
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n in decay_parameters and n not in projector_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": self.args.weight_decay,
+                    },
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n not in decay_parameters and n not in projector_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": 0.0,
+                    },
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n in decay_parameters and n in projector_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": self.args.weight_decay,
+                        "lr": self.args.vision2text_lr,
+                    },
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n not in decay_parameters and n in projector_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": 0.0,
+                        "lr": self.args.vision2text_lr,
+                    },
+                ]
+            else:
+                optimizer_grouped_parameters = [
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n in decay_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": self.args.weight_decay,
+                    },
+                    {
+                        "params": [
+                            p for n, p in opt_model.named_parameters() if (n not in decay_parameters and p.requires_grad)
+                        ],
+                        "weight_decay": 0.0,
+                    },
+                ]
+            ic(len(optimizer_grouped_parameters[0]['params']),len(optimizer_grouped_parameters[1]['params']))
+            optimizer_cls, optimizer_kwargs = Trainer.get_optimizer_cls_and_kwargs(self.args)
+
+            if self.sharded_ddp == ShardedDDPOption.SIMPLE:
+                self.optimizer = OSS(
+                    params=optimizer_grouped_parameters,
+                    optim=optimizer_cls,
+                    **optimizer_kwargs,
+                )
+            else:
+                self.optimizer = optimizer_cls(optimizer_grouped_parameters, **optimizer_kwargs)
+                if optimizer_cls.__name__ == "Adam8bit":
+                    import bitsandbytes
+
+                    manager = bitsandbytes.optim.GlobalOptimManager.get_instance()
+
+                    skipped = 0
+                    for module in opt_model.modules():
+                        if isinstance(module, nn.Embedding):
+                            skipped += sum({p.data_ptr(): p.numel() for p in module.parameters()}.values())
+                            logger.info(f"skipped {module}: {skipped/2**20}M params")
+                            manager.register_module_override(module, "weight", {"optim_bits": 32})
+                            logger.debug(f"bitsandbytes: will optimize {module} in fp32")
+                    logger.info(f"skipped: {skipped/2**20}M params")
+
+        return self.optimizer
+
+    def _save_checkpoint(self, model, trial, metrics=None):
+        super(MPLUGDocOwlTrainer, self)._save_checkpoint(model, trial, metrics)
+
+    def _save(self, output_dir: Optional[str] = None, state_dict=None):
+        super(MPLUGDocOwlTrainer, self)._save(output_dir, state_dict)
\ No newline at end of file

mplug_docowl/train/train_docowl.py

+# Adopted from https://github.com/lm-sys/FastChat. Below is the original copyright:
+# Adopted from tatsu-lab@stanford_alpaca. Below is the original copyright:
+#    Copyright 2023 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li
+#
+#    Licensed under the Apache License, Version 2.0 (the "License");
+#    you may not use this file except in compliance with the License.
+#    You may obtain a copy of the License at
+#
+#        http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS,
+#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#    See the License for the specific language governing permissions and
+#    limitations under the License.
+
+
+# Make it more memory efficient by monkey patching the LLaMA model with FlashAttn.
+
+## if you fail to import mplug_docowl, try adding absolute path of the mPLUG-DocOwl1.5 directory
+import os
+import copy
+from dataclasses import dataclass, field
+import json
+import jsonlines
+import logging
+import pathlib
+from typing import Dict, Optional, Sequence, List
+
+import torch
+
+import transformers
+from transformers.models.clip.image_processing_clip import CLIPImageProcessor
+
+from torch.utils.data import Dataset
+
+from PIL import Image
+from icecream import ic
+
+import sys
+sys.path.append('/home/wanglch/projects/mPLUG-DocOwl1.5-Omni') 
+print(sys.path)
+
+# Need to call this before importing transformers.
+from mplug_docowl.train.llama_flash_attn_monkey_patch import replace_llama_attn_with_flash_attn
+replace_llama_attn_with_flash_attn()
+
+from mplug_docowl.train.mplug_docowl_trainer import MPLUGDocOwlTrainer
+from mplug_docowl.constants import IGNORE_INDEX, IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN
+
+from mplug_docowl import conversation as conversation_lib
+from mplug_docowl.model import *
+from mplug_docowl.mm_utils import tokenizer_image_token
+from mplug_docowl.processor import DocProcessor
+
+local_rank = None
+
+
+def read_jsonl(filename):
+    lines = []
+    with open(filename, 'r', encoding='utf-8') as f:
+        for line in jsonlines.Reader(f):
+            lines.append(line)
+    return lines
+
+
+def rank0_print(*args):
+    if local_rank == 0:
+        print(*args)
+
+
+@dataclass
+class ModelArguments:
+    model_name_or_path: Optional[str] = field(default="facebook/opt-125m")
+    version: Optional[str] = field(default="v0")
+    freeze_backbone: bool = field(default=False)
+
+@dataclass
+class DataArguments:
+    data_path: str = field(default=None,
+                           metadata={"help": "Path to the training data."})
+    lazy_preprocess: bool = False
+    is_multimodal: bool = False
+    image_folder: Optional[str] = field(default=None)
+    image_aspect_ratio: str = 'square'
+    image_grid_pinpoints: Optional[str] = field(default=None)
+    image_size: int = 448
+    crop_anchors: str = 'grid_9'
+    add_global_img: bool = True
+    add_textual_crop_indicator: bool = True
+
+
+@dataclass
+class TrainingArguments(transformers.TrainingArguments):
+    cache_dir: Optional[str] = field(default=None)
+    optim: str = field(default="adamw_torch")
+    remove_unused_columns: bool = field(default=False)
+    
+    tune_vision2text: bool = field(default=True)
+    freeze_vision_model: bool = field(default=True)
+
+    model_max_length: int = field(
+        default=512,
+        metadata={
+            "help":
+            "Maximum sequence length. Sequences will be right padded (and possibly truncated)."
+        },
+    )
+    double_quant: bool = field(
+        default=True,
+        metadata={"help": "Compress the quantization statistics through double quantization."}
+    )
+    quant_type: str = field(
+        default="nf4",
+        metadata={"help": "Quantization data type to use. Should be one of `fp4` or `nf4`."}
+    )
+    bits: int = field(
+        default=16,
+        metadata={"help": "How many bits to use."}
+    )
+    lora_enable: bool = False
+    lora_r: int = 64
+    lora_alpha: int = 16
+    lora_dropout: float = 0.05
+    lora_weight_path: str = ""
+    lora_bias: str = "none"
+    vision2text_lr: Optional[float] = None
+    group_by_modality_length: bool = field(default=False)
+
+
+def maybe_zero_3(param, ignore_status=False, name=None):
+    from deepspeed import zero
+    from deepspeed.runtime.zero.partition_parameters import ZeroParamStatus
+    if hasattr(param, "ds_id"):
+        if param.ds_status == ZeroParamStatus.NOT_AVAILABLE:
+            if not ignore_status:
+                logging.warning(f"{name}: param.ds_status != ZeroParamStatus.NOT_AVAILABLE: {param.ds_status}")
+        with zero.GatheredParameters([param]):
+            param = param.data.detach().cpu().clone()
+    else:
+        param = param.detach().cpu().clone()
+    return param
+
+
+# Borrowed from peft.utils.get_peft_model_state_dict
+def get_peft_state_maybe_zero_3(named_params, bias):
+    if bias == "none":
+        to_return = {k: t for k, t in named_params if "lora_" in k}
+    elif bias == "all":
+        to_return = {k: t for k, t in named_params if "lora_" in k or "bias" in k}
+    elif bias == "lora_only":
+        to_return = {}
+        maybe_lora_bias = {}
+        lora_bias_names = set()
+        for k, t in named_params:
+            if "lora_" in k:
+                to_return[k] = t
+                bias_name = k.split("lora_")[0] + "bias"
+                lora_bias_names.add(bias_name)
+            elif "bias" in k:
+                maybe_lora_bias[k] = t
+        for k, t in maybe_lora_bias:
+            if bias_name in lora_bias_names:
+                to_return[bias_name] = t
+    else:
+        raise NotImplementedError
+    to_return = {k: maybe_zero_3(v, ignore_status=True) for k, v in to_return.items()}
+    return to_return
+
+
+def get_peft_state_non_lora_maybe_zero_3(named_params, require_grad_only=True):
+    to_return = {k: t for k, t in named_params if "lora_" not in k}
+    if require_grad_only:
+        to_return = {k: t for k, t in to_return.items() if t.requires_grad}
+    to_return = {k: maybe_zero_3(v, ignore_status=True).cpu() for k, v in to_return.items()}
+    return to_return
+
+
+def get_mm_adapter_state_maybe_zero_3(named_params, keys_to_match):
+    to_return = {k: t for k, t in named_params if any(key_match in k for key_match in keys_to_match)}
+    to_return = {k: maybe_zero_3(v, ignore_status=True).cpu() for k, v in to_return.items()}
+    return to_return
+
+
+def find_all_linear_names(model):
+    cls = torch.nn.Linear
+    lora_module_names = set()
+    multimodal_keywords = ['vision_model', 'vision2text']
+    for name, module in model.named_modules():
+        if any(mm_keyword in name for mm_keyword in multimodal_keywords):
+            continue
+        if isinstance(module, cls):
+            lora_module_names.add(name)
+
+    if 'lm_head' in lora_module_names: # needed for 16-bit
+        lora_module_names.remove('lm_head')
+    return list(lora_module_names)
+
+
+def safe_save_model_for_hf_trainer(trainer: transformers.Trainer,
+                                   output_dir: str):
+    """Collects the state dict and dump to disk."""
+
+    if trainer.deepspeed:
+        torch.cuda.synchronize()
+        trainer.save_model(output_dir)
+        return
+
+    state_dict = trainer.model.state_dict()
+    if trainer.args.should_save:
+        cpu_state_dict = {
+            key: value.cpu()
+            for key, value in state_dict.items()
+        }
+        del state_dict
+        trainer._save(output_dir, state_dict=cpu_state_dict)  # noqa
+
+
+def smart_tokenizer_and_embedding_resize(
+    special_tokens_dict: Dict,
+    tokenizer: transformers.PreTrainedTokenizer,
+    model: transformers.PreTrainedModel,
+):
+    """Resize tokenizer and embedding.
+
+    Note: This is the unoptimized version that may make your embedding size not be divisible by 64.
+    """
+    num_new_tokens = tokenizer.add_special_tokens(special_tokens_dict)
+    model.resize_token_embeddings(len(tokenizer))
+
+    if num_new_tokens > 0:
+        input_embeddings = model.get_input_embeddings().weight.data
+        output_embeddings = model.get_output_embeddings().weight.data
+
+        input_embeddings_avg = input_embeddings[:-num_new_tokens].mean(
+            dim=0, keepdim=True)
+        output_embeddings_avg = output_embeddings[:-num_new_tokens].mean(
+            dim=0, keepdim=True)
+
+        input_embeddings[-num_new_tokens:] = input_embeddings_avg
+        output_embeddings[-num_new_tokens:] = output_embeddings_avg
+
+
+def _tokenize_fn(strings: Sequence[str],
+                 tokenizer: transformers.PreTrainedTokenizer) -> Dict:
+    """Tokenize a list of strings."""
+    tokenized_list = [
+        tokenizer(
+            text,
+            return_tensors="pt",
+            padding="longest",
+            max_length=tokenizer.model_max_length,
+            truncation=True,
+        ) for text in strings
+    ]
+    input_ids = labels = [
+        tokenized.input_ids[0] for tokenized in tokenized_list
+    ]
+    input_ids_lens = labels_lens = [
+        tokenized.input_ids.ne(tokenizer.pad_token_id).sum().item()
+        for tokenized in tokenized_list
+    ]
+    return dict(
+        input_ids=input_ids,
+        labels=labels,
+        input_ids_lens=input_ids_lens,
+        labels_lens=labels_lens,
+    )
+
+
+def _mask_targets(target, tokenized_lens, speakers):
+    # cur_idx = 0
+    cur_idx = tokenized_lens[0]
+    tokenized_lens = tokenized_lens[1:]
+    target[:cur_idx] = IGNORE_INDEX
+    for tokenized_len, speaker in zip(tokenized_lens, speakers):
+        if speaker == "human":
+            target[cur_idx+2:cur_idx + tokenized_len] = IGNORE_INDEX
+        cur_idx += tokenized_len
+
+
+def _add_speaker_and_signal(header, source, get_conversation=True):
+    """Add speaker and start/end signal on each round."""
+    BEGIN_SIGNAL = "### "
+    END_SIGNAL = "\n"
+    conversation = header
+    for sentence in source:
+        from_str = sentence["from"]
+        if from_str.lower() == "human":
+            from_str = conversation_lib.default_conversation.roles[0]
+        elif from_str.lower() == "gpt":
+            from_str = conversation_lib.default_conversation.roles[1]
+        else:
+            from_str = 'unknown'
+        sentence["value"] = (BEGIN_SIGNAL + from_str + ": " +
+                             sentence["value"] + END_SIGNAL)
+        if get_conversation:
+            conversation += sentence["value"]
+    conversation += BEGIN_SIGNAL
+    return conversation
+
+
+"""def preprocess_multimodal(
+    sources: Sequence[str],
+    data_args: DataArguments
+) -> Dict:
+    is_multimodal = data_args.is_multimodal
+    if not is_multimodal:
+        return sources
+
+    for source in sources:
+        for sentence in source:
+            if DEFAULT_IMAGE_TOKEN in sentence['value']:
+                sentence['value'] = sentence['value'].replace(DEFAULT_IMAGE_TOKEN, '').strip()
+                sentence['value'] = DEFAULT_IMAGE_TOKEN + '\n' + sentence['value']
+                sentence['value'] = sentence['value'].strip()
+               
+            replace_token = DEFAULT_IMAGE_TOKEN
+            sentence["value"] = sentence["value"].replace(DEFAULT_IMAGE_TOKEN, replace_token)
+
+    return sources"""
+
+
+def docowl_text_preprocess_v1(
+    source,
+    tokenizer: transformers.PreTrainedTokenizer,
+    has_image: bool = False
+) -> Dict:
+    """
+    source: list of {'role':'user'/'assistant', 'content':xxxx}
+    """
+    conv = conversation_lib.default_conversation.copy()
+    # conv.roles: ("USER", "ASSISTANT")
+    roles = {"user": conv.roles[0], "assistant": conv.roles[1]}
+
+
+    # Apply prompt templates
+    conversations = []
+
+    # Skip the first one if it is not from human
+    if roles[source[0]["role"]] != conv.roles[0]:
+        source = source[1:]
+
+    conv.messages = []
+    for j, sentence in enumerate(source):
+        role = roles[sentence["role"]]
+        assert role == conv.roles[j % 2]
+        conv.append_message(role, sentence["content"])
+    
+    # conv.get_prompt(): USER: {content} ASSISTANT: {content}</s>USER: {content} ASSISTANT: {content}</s>...
+    conversations.append(conv.get_prompt()) 
+
+    # Tokenize conversations
+    if has_image:
+        input_ids = torch.stack([tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations], dim=0)
+    else:
+        input_ids = tokenizer(
+            conversations,
+            return_tensors="pt",
+            padding="longest",
+            max_length=tokenizer.model_max_length,
+            truncation=True,
+        ).input_ids
+
+    targets = input_ids.clone()
+
+    assert conv.sep_style == conversation_lib.SeparatorStyle.TWO or conv.sep_style == conversation_lib.SeparatorStyle.TWO_NO_SYS
+
+    # Mask targets
+    sep = conv.sep + conv.roles[1] + ": " # ' ASSISTANT: '
+    for conversation, target in zip(conversations, targets):
+        total_len = int(target.ne(tokenizer.pad_token_id).sum())
+
+        rounds = conversation.split(conv.sep2) # split by </s>
+        cur_len = 1
+        target[:cur_len] = IGNORE_INDEX
+        for i, rou in enumerate(rounds):
+            if rou == "":
+                break
+
+            parts = rou.split(sep) # split each round by ' ASSISTANT: '
+            if len(parts) != 2:
+                break
+            parts[0] += sep # input query, ignore for loss
+
+            if has_image:
+                round_len = len(tokenizer_image_token(rou, tokenizer))
+                instruction_len = len(tokenizer_image_token(parts[0], tokenizer)) - 2
+            else:
+                round_len = len(tokenizer(rou).input_ids)
+                instruction_len = len(tokenizer(parts[0]).input_ids) - 2
+
+            target[cur_len : cur_len + instruction_len] = IGNORE_INDEX
+
+            cur_len += round_len
+        target[cur_len:] = IGNORE_INDEX
+
+        if cur_len < tokenizer.model_max_length: # ignore padding
+            if cur_len != total_len:
+                target[:] = IGNORE_INDEX
+                print(
+                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
+                    f" (ignored)"
+                )
+
+    return dict(
+        input_ids=input_ids,
+        labels=targets,
+    )
+
+
+def preprocess_plain(
+    sources: Sequence[str],
+    tokenizer: transformers.PreTrainedTokenizer,
+) -> Dict:
+    # add end signal and concatenate together
+    conversations = []
+    for source in sources:
+        assert len(source) == 2
+        assert DEFAULT_IMAGE_TOKEN in source[0]['value']
+        source[0]['value'] = DEFAULT_IMAGE_TOKEN
+        conversation = source[0]['value'] + source[1]['value'] + conversation_lib.default_conversation.sep
+        conversations.append(conversation)
+    # tokenize conversations
+    input_ids = [tokenizer_image_token(prompt, tokenizer, return_tensors='pt') for prompt in conversations]
+    targets = copy.deepcopy(input_ids)
+    for target, source in zip(targets, sources):
+        tokenized_len = len(tokenizer_image_token(source[0]['value'], tokenizer))
+        target[:tokenized_len] = IGNORE_INDEX
+
+    return dict(input_ids=input_ids, labels=targets)
+
+class LazySupervisedDataset(Dataset):
+    """Dataset for supervised fine-tuning."""
+
+    def __init__(self, data_path: str,
+                 tokenizer: transformers.PreTrainedTokenizer,
+                 data_args: DataArguments):
+        super(LazySupervisedDataset, self).__init__()
+
+        list_data_dict = read_jsonl(data_path)
+
+        rank0_print("Formatting inputs...Skip in lazy mode")
+        self.tokenizer = tokenizer
+        self.list_data_dict = list_data_dict
+        self.data_args = data_args
+        self.doc_image_processor = DocProcessor(image_size=self.data_args.image_size, 
+                                                anchors=self.data_args.crop_anchors, 
+                                                add_global_img=self.data_args.add_global_img, 
+                                                add_textual_crop_indicator=self.data_args.add_textual_crop_indicator)
+
+    def __len__(self):
+        return len(self.list_data_dict)
+    
+    """@property
+    def lengths(self):
+        length_list = []
+        for sample in self.list_data_dict:
+            img_tokens = 128 if 'image' in sample else 0
+            length_list.append(sum(len(conv['value'].split()) for conv in sample['conversations']) + img_tokens)
+        return length_list"""
+
+    """@property
+    def modality_lengths(self):
+        length_list = []
+        for sample in self.list_data_dict:
+            cur_len = sum(len(conv['value'].split()) for conv in sample['conversations'])
+            cur_len = cur_len if 'image' in sample else -cur_len
+            length_list.append(cur_len)
+        return length_list"""
+
+    def next_rand(self):
+        import random
+        return random.randint(0,len(self)-1)
+
+    def __getitem__(self, i) -> Dict[str, torch.Tensor]:
+        while True:
+            source = self.list_data_dict[i]
+            assert 'image' in source
+            image_file = source['image'][0]
+            image_folder = self.data_args.image_folder
+            from pathlib import Path
+            if not Path(os.path.join(image_folder, image_file)).exists():
+                i = self.next_rand()
+                continue
+            ## crop image and revise query acccording to the cropping shape
+            image_path=os.path.join(image_folder, image_file)
+            query = source['messages'][0]['content'] # e.g. <|image|>what is the contact person name mentioned in letter?."
+            # image: tensor, [N_crops+1, c, h, w]
+            # patch_positions: tensor, [N_crops+1, 2]
+            # text: string, add textual crop indicators and repeat the <|image|> N_crops+1 times
+            image, patch_positions, processed_query = self.doc_image_processor(images=image_path, query=query)
+            source['messages'][0]['content'] = processed_query
+
+            ## text tokenization
+            data_dict = docowl_text_preprocess_v1(
+                source['messages'],
+                self.tokenizer,
+                has_image=('image' in self.list_data_dict[i]))
+            if isinstance(i, int):
+                data_dict = dict(input_ids=data_dict["input_ids"][0],
+                                labels=data_dict["labels"][0])
+
+            # image exist in the data
+            data_dict['image'] = image
+            data_dict['patch_positions'] = patch_positions
+            
+            
+            return data_dict
+
+
+@dataclass
+class DataCollatorForSupervisedDataset(object):
+    """Collate examples for supervised fine-tuning."""
+
+    tokenizer: transformers.PreTrainedTokenizer
+
+    def __call__(self, instances: Sequence[Dict]) -> Dict[str, torch.Tensor]:
+        input_ids, labels = tuple([instance[key] for instance in instances]
+                                  for key in ("input_ids", "labels"))
+        input_ids = torch.nn.utils.rnn.pad_sequence(
+            input_ids,
+            batch_first=True,
+            padding_value=self.tokenizer.pad_token_id)
+        labels = torch.nn.utils.rnn.pad_sequence(labels,
+                                                 batch_first=True,
+                                                 padding_value=IGNORE_INDEX)
+        input_ids = input_ids[:, :self.tokenizer.model_max_length]
+        labels = labels[:, :self.tokenizer.model_max_length]
+        batch = dict(
+            input_ids=input_ids,
+            labels=labels,
+            attention_mask=input_ids.ne(self.tokenizer.pad_token_id),
+        )
+
+        if 'image' in instances[0]:
+            images = [instance['image'] for instance in instances]
+            batch['images'] = torch.cat(images) # Sum(Crop+1) x c x h x w
+        
+        if 'patch_positions' in instances[0]:
+            patch_positions = [instance['patch_positions'] for instance in instances]
+            batch['patch_positions'] = torch.cat(patch_positions) # Sum(Crop+1) x 2
+
+        return batch
+
+
+def make_supervised_data_module(tokenizer: transformers.PreTrainedTokenizer,
+                                data_args) -> Dict:
+    """Make dataset and collator for supervised fine-tuning."""
+    train_dataset = LazySupervisedDataset(tokenizer=tokenizer,
+                                data_path=data_args.data_path,
+                                data_args=data_args)
+    data_collator = DataCollatorForSupervisedDataset(tokenizer=tokenizer)
+    return dict(train_dataset=train_dataset,
+                eval_dataset=None,
+                data_collator=data_collator)
+
+
+def train():
+    global local_rank
+
+    parser = transformers.HfArgumentParser(
+        (ModelArguments, DataArguments, TrainingArguments))
+    model_args, data_args, training_args = parser.parse_args_into_dataclasses()
+    local_rank = training_args.local_rank
+    compute_dtype = (torch.float16 if training_args.fp16 else (torch.bfloat16 if training_args.bf16 else torch.float32))
+
+    bnb_model_from_pretrained_args = {}
+    if training_args.bits in [4, 8]:
+        from transformers import BitsAndBytesConfig
+        bnb_model_from_pretrained_args.update(dict(
+            device_map={"": training_args.device},
+            load_in_4bit=training_args.bits == 4,
+            load_in_8bit=training_args.bits == 8,
+            quantization_config=BitsAndBytesConfig(
+                load_in_4bit=training_args.bits == 4,
+                load_in_8bit=training_args.bits == 8,
+                llm_int8_threshold=6.0,
+                llm_int8_has_fp16_weight=False,
+                bnb_4bit_compute_dtype=compute_dtype,
+                bnb_4bit_use_double_quant=training_args.double_quant,
+                bnb_4bit_quant_type=training_args.quant_type # {'fp4', 'nf4'}
+            )
+        ))
+
+    # 
+    model = MPLUGDocOwlLlamaForCausalLM.from_pretrained(
+        model_args.model_name_or_path,
+        cache_dir=training_args.cache_dir,
+        **bnb_model_from_pretrained_args
+    )
+    model.config.use_cache = False
+
+    ic(model_args.freeze_backbone)
+    if model_args.freeze_backbone:
+        model.model.requires_grad_(False)
+
+    if training_args.bits in [4, 8]:
+        from peft import prepare_model_for_kbit_training
+        model.config.torch_dtype=(torch.float32 if training_args.fp16 else (torch.bfloat16 if training_args.bf16 else torch.float32))
+        model = prepare_model_for_kbit_training(model, use_gradient_checkpointing=training_args.gradient_checkpointing)
+
+    if training_args.gradient_checkpointing:
+        if hasattr(model, "enable_input_require_grads"): # default true
+            model.enable_input_require_grads()
+            # print('model.enable_input_require_grads()')
+        else:
+            def make_inputs_require_grad(module, input, output):
+                output.requires_grad_(True)
+            model.get_input_embeddings().register_forward_hook(make_inputs_require_grad)
+
+    if training_args.lora_enable:
+        from peft import LoraConfig, get_peft_model
+        lora_config = LoraConfig(
+            r=training_args.lora_r,
+            lora_alpha=training_args.lora_alpha,
+            target_modules=find_all_linear_names(model),
+            lora_dropout=training_args.lora_dropout,
+            bias=training_args.lora_bias,
+            task_type="CAUSAL_LM",
+        )
+        if training_args.bits == 16:
+            if training_args.bf16:
+                model.to(torch.bfloat16)
+            if training_args.fp16:
+                model.to(torch.float16)
+        rank0_print("Adding LoRA adapters...")
+        model = get_peft_model(model, lora_config)
+
+    tokenizer = transformers.AutoTokenizer.from_pretrained(
+        model_args.model_name_or_path,
+        cache_dir=training_args.cache_dir,
+        model_max_length=training_args.model_max_length,
+        padding_side="right",
+        use_fast=False,
+    )
+
+    
+    tokenizer.pad_token = tokenizer.unk_token
+    if model_args.version in conversation_lib.conv_templates:
+        conversation_lib.default_conversation = conversation_lib.conv_templates[model_args.version]
+    else:
+        conversation_lib.default_conversation = conversation_lib.conv_templates["vicuna_v1"]
+    
+    if not training_args.freeze_vision_model and training_args.bits in [4, 8]:
+        model.get_model().vision_model.to(dtype=compute_dtype, device=training_args.device)
+    else:
+        vision_tower = model.get_model().vision_model
+        vision_tower.to(dtype=torch.bfloat16 if training_args.bf16 else torch.float16, device=training_args.device)
+    
+    if training_args.tune_vision2text and training_args.bits in [4, 8]:
+        model.get_model().vision2text.to(dtype=compute_dtype, device=training_args.device)
+    else:
+        vision2text = model.get_model().vision2text
+        vision2text.to(dtype=torch.bfloat16 if training_args.bf16 else torch.float16, device=training_args.device)
+    
+    data_args.is_multimodal = True
+    model.config.image_aspect_ratio = data_args.image_aspect_ratio
+    model.config.image_grid_pinpoints = data_args.image_grid_pinpoints
+    model.config.tune_vision2text = model_args.tune_vision2text = training_args.tune_vision2text
+    ic(training_args.tune_vision2text)
+    # model.requires_grad_(True)
+    if training_args.tune_vision2text:
+        # model.requires_grad_(False)
+        for p in model.get_model().vision2text.parameters():
+            p.requires_grad = True
+            
+    model.config.freeze_vision_model = training_args.freeze_vision_model
+    ic(training_args.freeze_vision_model)
+    if training_args.freeze_vision_model:
+        for p in model.get_model().vision_model.parameters():
+            p.requires_grad = False
+            
+    model.config.vision2text_lr = training_args.vision2text_lr
+
+
+    if training_args.bits in [4, 8]:
+        from peft.tuners.lora import LoraLayer
+        for name, module in model.named_modules():
+            if isinstance(module, LoraLayer):
+                if training_args.bf16:
+                    module = module.to(torch.bfloat16)
+            if 'norm' in name:
+                module = module.to(torch.float32)
+            if 'lm_head' in name or 'embed_tokens' in name:
+                if hasattr(module, 'weight'):
+                    if training_args.bf16 and module.weight.dtype == torch.float32:
+                        module = module.to(torch.bfloat16)
+
+
+    data_module = make_supervised_data_module(tokenizer=tokenizer,
+                                              data_args=data_args)
+
+    trainer = MPLUGDocOwlTrainer(model=model,
+                    tokenizer=tokenizer,
+                    args=training_args,
+                    **data_module)
+
+
+    trainer.train()
+
+    trainer.save_state()
+
+    model.config.use_cache = True
+
+    if training_args.lora_enable:
+        state_dict = get_peft_state_maybe_zero_3(
+            model.named_parameters(), training_args.lora_bias
+        )
+        non_lora_state_dict = get_peft_state_non_lora_maybe_zero_3(
+            model.named_parameters()
+        )
+        if training_args.local_rank == 0 or training_args.local_rank == -1:
+            model.config.save_pretrained(training_args.output_dir)
+            model.save_pretrained(training_args.output_dir, state_dict=state_dict)
+            torch.save(non_lora_state_dict, os.path.join(training_args.output_dir, 'non_lora_trainables.bin'))
+    else:
+        safe_save_model_for_hf_trainer(trainer=trainer,
+                                       output_dir=training_args.output_dir)
+
+
+if __name__ == "__main__":
+    train()
\ No newline at end of file