monkey_model

monkey_model/config.json

+{
+    "architectures": [
+      "MonkeyLMHeadModel"
+    ],
+    "attn_dropout_prob": 0.0,
+    "auto_map": {
+      "AutoConfig": "configuration_qwen.QWenConfig",
+      "AutoModelForCausalLM": "modeling_monkey.MonkeyLMHeadModel"
+    },
+    "bf16": true,
+    "emb_dropout_prob": 0.0,
+    "fp16": false,
+    "fp32": false,
+    "hidden_size": 4096,
+    "initializer_range": 0.02,
+    "intermediate_size": 22016,
+    "kv_channels": 128,
+    "layer_norm_epsilon": 1e-06,
+    "max_position_embeddings": 8192,
+    "model_type": "monkey",
+    "no_bias": true,
+    "num_attention_heads": 32,
+    "num_hidden_layers": 32,
+    "onnx_safe": null,
+    "rotary_emb_base": 10000,
+    "rotary_pct": 1.0,
+    "scale_attn_weights": true,
+    "seq_length": 2048,
+    "tie_word_embeddings": false,
+    "tokenizer_type": "QWenTokenizer",
+    "torch_dtype": "bfloat16",
+    "transformers_version": "4.32.0",
+    "use_cache": false,
+    "use_dynamic_ntk": true,
+    "use_flash_attn": false,
+    "use_logn_attn": true,
+    "visual": {
+      "heads": 16,
+      "image_size": 896,
+      "image_start_id": 151857,
+      "layers": 48,
+      "mlp_ratio": 4.9231,
+      "output_dim": 4096,
+      "patch_size": 14,
+      "width": 1664,
+      "lora_repeat_num":0
+    },
+    "vocab_size": 151936
+  }
+  
\ No newline at end of file

monkey_model/configuration_monkey.py

+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+from transformers import PretrainedConfig
+
+
+class MonkeyConfig(PretrainedConfig):
+    model_type = "monkey"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=151936,
+        hidden_size=4096,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        emb_dropout_prob=0.0,
+        attn_dropout_prob=0.0,
+        layer_norm_epsilon=1e-6,
+        initializer_range=0.02,
+        max_position_embeddings=8192,
+        scale_attn_weights=True,
+        use_cache=True,
+        bf16=False,
+        fp16=False,
+        fp32=False,
+        kv_channels=128,
+        rotary_pct=1.0,
+        rotary_emb_base=10000,
+        use_dynamic_ntk=True,
+        use_logn_attn=True,
+        use_flash_attn="auto",
+        intermediate_size=22016,
+        no_bias=True,
+        tie_word_embeddings=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.emb_dropout_prob = emb_dropout_prob
+        self.attn_dropout_prob = attn_dropout_prob
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.initializer_range = initializer_range
+        self.scale_attn_weights = scale_attn_weights
+        self.use_cache = use_cache
+        self.max_position_embeddings = max_position_embeddings
+        self.bf16 = bf16
+        self.fp16 = fp16
+        self.fp32 = fp32
+        self.kv_channels = kv_channels
+        self.rotary_pct = rotary_pct
+        self.rotary_emb_base = rotary_emb_base
+        self.use_dynamic_ntk = use_dynamic_ntk
+        self.use_logn_attn = use_logn_attn
+        self.use_flash_attn = use_flash_attn
+        self.no_bias = no_bias
+        super().__init__(
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs
+        )

monkey_model/configuration_qwen.py

+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+from transformers import PretrainedConfig
+
+
+class QwenConfig(PretrainedConfig):
+    model_type = "monkey"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=151936,
+        hidden_size=4096,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        emb_dropout_prob=0.0,
+        attn_dropout_prob=0.0,
+        layer_norm_epsilon=1e-6,
+        initializer_range=0.02,
+        max_position_embeddings=8192,
+        scale_attn_weights=True,
+        use_cache=True,
+        bf16=False,
+        fp16=False,
+        fp32=False,
+        kv_channels=128,
+        rotary_pct=1.0,
+        rotary_emb_base=10000,
+        use_dynamic_ntk=True,
+        use_logn_attn=True,
+        use_flash_attn="auto",
+        intermediate_size=22016,
+        no_bias=True,
+        tie_word_embeddings=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.emb_dropout_prob = emb_dropout_prob
+        self.attn_dropout_prob = attn_dropout_prob
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.initializer_range = initializer_range
+        self.scale_attn_weights = scale_attn_weights
+        self.use_cache = use_cache
+        self.max_position_embeddings = max_position_embeddings
+        self.bf16 = bf16
+        self.fp16 = fp16
+        self.fp32 = fp32
+        self.kv_channels = kv_channels
+        self.rotary_pct = rotary_pct
+        self.rotary_emb_base = rotary_emb_base
+        self.use_dynamic_ntk = use_dynamic_ntk
+        self.use_logn_attn = use_logn_attn
+        self.use_flash_attn = use_flash_attn
+        self.no_bias = no_bias
+        super().__init__(
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs
+        )

monkey_model/modeling_monkey.py

+import importlib
+import math
+from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List, Any, Generator
+
+import torch
+import torch.nn.functional as F
+import torch.utils.checkpoint
+from torch.cuda.amp import autocast
+
+from torch.nn import CrossEntropyLoss
+from transformers import PreTrainedTokenizer, GenerationConfig, StoppingCriteriaList
+from transformers.generation.logits_process import LogitsProcessorList
+
+if TYPE_CHECKING:
+    from transformers.generation.streamers import BaseStreamer
+from transformers.generation.utils import GenerateOutput
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training
+try:
+    from einops import rearrange
+except ImportError:
+    rearrange = None
+from torch import nn
+from monkey_model.modeling_qwen import QWenModel,QWenPreTrainedModel,QWenLMHeadModel
+SUPPORT_CUDA = torch.cuda.is_available()
+SUPPORT_BF16 = SUPPORT_CUDA and torch.cuda.is_bf16_supported()
+SUPPORT_FP16 = SUPPORT_CUDA and torch.cuda.get_device_capability(0)[0] >= 7
+logger = logging.get_logger(__name__)
+class MonkeyModel(QWenModel):
+    def __init__(self, config):
+        super().__init__(config)
+    
+    
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        token_type_ids: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: 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,
+    ):
+        if past_key_values is None and torch.any(input_ids == self.config.visual['image_start_id']):
+            bos_pos = torch.where(input_ids == self.config.visual['image_start_id'])
+            eos_pos = torch.where(input_ids == self.config.visual['image_start_id'] + 1)
+            assert (bos_pos[0] == eos_pos[0]).all()
+            img_pos = torch.stack((bos_pos[0], bos_pos[1], eos_pos[1]), dim=1)
+            images = []
+            for i, a, b in img_pos:
+                image = input_ids[i][a + 1 : b - 1].tolist()
+                image = image[ : image.index(self.config.visual['image_start_id'] + 2)]
+                images.append(bytes(image).decode('utf-8'))
+            windows,images_448 = self.visual.encode(images)
+            patch_list = []
+            lora_idx = 0 
+            for col in windows:
+                for image_patch in col:
+                    patch_list.append(self.visual(image_patch,idx=lora_idx))
+                    lora_idx += 1
+                    
+            global_feat = self.visual(images_448)
+            local_feat = torch.cat(patch_list,dim=1)
+            images = torch.cat([local_feat,global_feat],dim=1)
+            assert images.shape[0] == len(images)
+        else:
+            images = None
+        return super().forward(input_ids,
+            past_key_values,
+            attention_mask,
+            token_type_ids,
+            position_ids,
+            head_mask,inputs_embeds,
+            encoder_hidden_states,
+            encoder_attention_mask,
+            use_cache,
+            output_attentions,
+            output_hidden_states,
+            return_dict,
+            images)
+    
+
+
+
+class MonkeyLMHeadModel(QWenLMHeadModel):
+    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.rotary_emb\.inv_freq"]
+    _keys_to_ignore_on_load_unexpected = [r"h\.\d+\.attn\.masked_bias"]
+
+    def __init__(self, config):
+        super().__init__(config)
+        assert (
+            config.bf16 + config.fp16 + config.fp32 <= 1
+        ), "Only one of \"bf16\", \"fp16\", \"fp32\" can be true"
+
+        autoset_precision = config.bf16 + config.fp16 + config.fp32 == 0
+
+        if autoset_precision:
+            if SUPPORT_BF16:
+                logger.warn(
+                    "The model is automatically converting to bf16 for faster inference. "
+                    "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"."
+                )
+                config.bf16 = True
+            elif SUPPORT_FP16:
+                logger.warn(
+                    "The model is automatically converting to fp16 for faster inference. "
+                    "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"."
+                )
+                config.fp16 = True
+            else:
+                config.fp32 = True
+
+        if config.bf16 and SUPPORT_CUDA and not SUPPORT_BF16:
+            logger.warn("Your device does NOT seem to support bf16, you can switch to fp16 or fp32 by by passing fp16/fp32=True in \"AutoModelForCausalLM.from_pretrained\".")
+        if config.fp16 and SUPPORT_CUDA and not SUPPORT_FP16:
+            logger.warn("Your device does NOT support faster inference with fp16, please switch to fp32 which is likely to be faster")
+        if config.fp32:
+            if SUPPORT_BF16:
+                logger.warn("Your device support faster inference by passing bf16=True in \"AutoModelForCausalLM.from_pretrained\".")
+            elif SUPPORT_FP16:
+                logger.warn("Your device support faster inference by passing fp16=True in \"AutoModelForCausalLM.from_pretrained\".")
+
+        self.transformer = MonkeyModel(config)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        if config.bf16:
+            self.transformer.bfloat16()
+            self.lm_head.bfloat16()
+        if config.fp16:
+            self.transformer.half()
+            self.lm_head.half()
+        self.post_init()
+
+

monkey_model/qwen_generation_utils.py

+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Generation support."""
+
+from typing import Tuple, List, Union, Iterable
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+from transformers import PreTrainedTokenizer
+from transformers import logging
+from transformers.generation import LogitsProcessor
+
+logger = logging.get_logger(__name__)
+
+# Types.
+HistoryType = List[Tuple[str, str]]
+TokensType = List[int]
+BatchTokensType = List[List[int]]
+
+
+def pad_batch(batch: BatchTokensType, pad_id: int, seq_length: int) -> BatchTokensType:
+    for tokens in batch:
+        context_length = len(tokens)
+        if context_length < seq_length:
+            tokens.extend([pad_id] * (seq_length - context_length))
+    return batch
+
+
+def get_ltor_masks_and_position_ids(
+    data,
+    eod_token,
+    reset_position_ids,
+    reset_attention_mask,
+    eod_mask_loss,
+):
+    """Build masks and position id for left to right model."""
+
+    # Extract batch size and sequence length.
+    micro_batch_size, seq_length = data.size()
+
+    # Attention mask (lower triangular).
+    if reset_attention_mask:
+        att_mask_batch = micro_batch_size
+    else:
+        att_mask_batch = 1
+    attention_mask = torch.tril(
+        torch.ones((att_mask_batch, seq_length, seq_length), device=data.device)
+    ).view(att_mask_batch, 1, seq_length, seq_length)
+
+    # Loss mask.
+    loss_mask = torch.ones(data.size(), dtype=torch.float, device=data.device)
+    if eod_mask_loss:
+        loss_mask[data == eod_token] = 0.0
+
+    # Position ids.
+    position_ids = torch.arange(seq_length, dtype=torch.long, device=data.device)
+    position_ids = position_ids.unsqueeze(0).expand_as(data)
+    # We need to clone as the ids will be modifed based on batch index.
+    if reset_position_ids:
+        position_ids = position_ids.clone()
+
+    if reset_position_ids or reset_attention_mask:
+        # Loop through the batches:
+        for b in range(micro_batch_size):
+
+            # Find indecies where EOD token is.
+            eod_index = position_ids[b, data[b] == eod_token]
+            # Detach indecies from positions if going to modify positions.
+            if reset_position_ids:
+                eod_index = eod_index.clone()
+
+            # Loop through EOD indecies:
+            prev_index = 0
+            for j in range(eod_index.size()[0]):
+                i = eod_index[j]
+                # Mask attention loss.
+                if reset_attention_mask:
+                    attention_mask[b, 0, (i + 1) :, : (i + 1)] = 0
+                # Reset positions.
+                if reset_position_ids:
+                    position_ids[b, (i + 1) :] -= i + 1 - prev_index
+                    prev_index = i + 1
+
+    # Convert attention mask to binary:
+    attention_mask = attention_mask < 0.5
+
+    return attention_mask, loss_mask, position_ids
+
+
+def get_batch(context_tokens: torch.LongTensor, eod_id: int):
+    """Generate batch from context tokens."""
+    # Move to GPU.
+    tokens = context_tokens.contiguous().to(context_tokens.device)
+    # Get the attention mask and postition ids.
+    attention_mask, _, position_ids = get_ltor_masks_and_position_ids(
+        tokens,
+        eod_id,
+        reset_position_ids=False,
+        reset_attention_mask=False,
+        eod_mask_loss=False,
+    )
+    return tokens, attention_mask, position_ids
+
+
+def get_stop_words_ids(chat_format, tokenizer):
+    if chat_format == "raw":
+        stop_words_ids = [tokenizer.encode("Human:"), [tokenizer.eod_id]]
+    elif chat_format == "chatml":
+        stop_words_ids = [[tokenizer.im_end_id], [tokenizer.im_start_id]]
+    else:
+        raise NotImplementedError(f"Unknown chat format {chat_format!r}")
+    return stop_words_ids
+
+
+def make_context(
+    tokenizer: PreTrainedTokenizer,
+    query: str,
+    history: List[Tuple[str, str]] = None,
+    system: str = "",
+    max_window_size: int = 6144,
+    chat_format: str = "chatml",
+):
+    if history is None:
+        history = []
+
+    if chat_format == "chatml":
+        im_start, im_end = "<|im_start|>", "<|im_end|>"
+        im_start_tokens = [tokenizer.im_start_id]
+        im_end_tokens = [tokenizer.im_end_id]
+        nl_tokens = tokenizer.encode("\n")
+
+        def _tokenize_str(role, content):
+            return f"{role}\n{content}", tokenizer.encode(
+                role, allowed_special=set(tokenizer.IMAGE_ST)
+            ) + nl_tokens + tokenizer.encode(content, allowed_special=set(tokenizer.IMAGE_ST))
+
+        system_text, system_tokens_part = _tokenize_str("system", system)
+        system_tokens = im_start_tokens + system_tokens_part + im_end_tokens
+
+        raw_text = ""
+        context_tokens = []
+
+        for turn_query, turn_response in reversed(history):
+            query_text, query_tokens_part = _tokenize_str("user", turn_query)
+            query_tokens = im_start_tokens + query_tokens_part + im_end_tokens
+            if turn_response is not None:
+                response_text, response_tokens_part = _tokenize_str(
+                    "assistant", turn_response
+                )
+                response_tokens = im_start_tokens + response_tokens_part + im_end_tokens
+
+                next_context_tokens = nl_tokens + query_tokens + nl_tokens + response_tokens
+                prev_chat = (
+                    f"\n{im_start}{query_text}{im_end}\n{im_start}{response_text}{im_end}"
+                )
+            else:
+                next_context_tokens = nl_tokens + query_tokens + nl_tokens
+                prev_chat = f"\n{im_start}{query_text}{im_end}\n"
+
+            current_context_size = (
+                len(system_tokens) + len(next_context_tokens) + len(context_tokens)
+            )
+            if current_context_size < max_window_size:
+                context_tokens = next_context_tokens + context_tokens
+                raw_text = prev_chat + raw_text
+            else:
+                break
+
+        context_tokens = system_tokens + context_tokens
+        raw_text = f"{im_start}{system_text}{im_end}" + raw_text
+        context_tokens += (
+            nl_tokens
+            + im_start_tokens
+            + _tokenize_str("user", query)[1]
+            + im_end_tokens
+            + nl_tokens
+            + im_start_tokens
+            + tokenizer.encode("assistant")
+            + nl_tokens
+        )
+        raw_text += f"\n{im_start}user\n{query}{im_end}\n{im_start}assistant\n"
+
+    elif chat_format == "raw":
+        raw_text = query
+        context_tokens = tokenizer.encode(raw_text)
+    else:
+        raise NotImplementedError(f"Unknown chat format {chat_format!r}")
+
+    return raw_text, context_tokens
+
+
+def _decode_default(
+    tokens: List[int],
+    *,
+    stop_words: List[str],
+    eod_words: List[str],
+    tokenizer: PreTrainedTokenizer,
+    raw_text_len: int,
+    verbose: bool = False,
+    return_end_reason: bool = False,
+    errors: str='replace',
+):
+    trim_decode_tokens = tokenizer.decode(tokens, errors=errors)[raw_text_len:]
+    if verbose:
+        print("\nRaw Generate: ", trim_decode_tokens)
+
+    end_reason = f"Gen length {len(tokens)}"
+    for stop_word in stop_words:
+        trim_decode_tokens = trim_decode_tokens.replace(stop_word, "").strip()
+    for eod_word in eod_words:
+        if eod_word in trim_decode_tokens:
+            end_reason = f"Gen {eod_word!r}"
+        trim_decode_tokens = trim_decode_tokens.split(eod_word)[0]
+    trim_decode_tokens = trim_decode_tokens.strip()
+    if verbose:
+        print("\nEnd Reason:", end_reason)
+        print("\nGenerate: ", trim_decode_tokens)
+
+    if return_end_reason:
+        return trim_decode_tokens, end_reason
+    else:
+        return trim_decode_tokens
+
+
+def _decode_chatml(
+    tokens: List[int],
+    *,
+    stop_words: List[str],
+    eod_token_ids: List[int],
+    tokenizer: PreTrainedTokenizer,
+    raw_text_len: int,
+    context_length: int,
+    verbose: bool = False,
+    return_end_reason: bool = False,
+    errors: str='replace'
+):
+    end_reason = f"Gen length {len(tokens)}"
+    eod_token_idx = context_length
+    for eod_token_idx in range(context_length, len(tokens)):
+        if tokens[eod_token_idx] in eod_token_ids:
+            end_reason = f"Gen {tokenizer.decode([tokens[eod_token_idx]])!r}"
+            break
+
+    trim_decode_tokens = tokenizer.decode(tokens[:eod_token_idx], errors=errors)[raw_text_len:]
+    if verbose:
+        print("\nRaw Generate w/o EOD:", tokenizer.decode(tokens, errors=errors)[raw_text_len:])
+        print("\nRaw Generate:", trim_decode_tokens)
+        print("\nEnd Reason:", end_reason)
+    for stop_word in stop_words:
+        trim_decode_tokens = trim_decode_tokens.replace(stop_word, "").strip()
+    trim_decode_tokens = trim_decode_tokens.strip()
+    if verbose:
+        print("\nGenerate:", trim_decode_tokens)
+
+    if return_end_reason:
+        return trim_decode_tokens, end_reason
+    else:
+        return trim_decode_tokens
+
+
+def decode_tokens(
+    tokens: Union[torch.LongTensor, TokensType],
+    tokenizer: PreTrainedTokenizer,
+    raw_text_len: int,
+    context_length: int,
+    chat_format: str,
+    verbose: bool = False,
+    return_end_reason: bool = False,
+    errors: str="replace",
+) -> str:
+    if torch.is_tensor(tokens):
+        tokens = tokens.cpu().numpy().tolist()
+
+    if chat_format == "chatml":
+        return _decode_chatml(
+            tokens,
+            stop_words=[],
+            eod_token_ids=[tokenizer.im_start_id, tokenizer.im_end_id],
+            tokenizer=tokenizer,
+            raw_text_len=raw_text_len,
+            context_length=context_length,
+            verbose=verbose,
+            return_end_reason=return_end_reason,
+            errors=errors,
+        )
+    elif chat_format == "raw":
+        return _decode_default(
+            tokens,
+            stop_words=["<|endoftext|>"],
+            eod_words=["<|endoftext|>"],
+            tokenizer=tokenizer,
+            raw_text_len=raw_text_len,
+            verbose=verbose,
+            return_end_reason=return_end_reason,
+            errors=errors,
+        )
+    else:
+        raise NotImplementedError(f"Unknown chat format {chat_format!r}")
+
+
+class StopWordsLogitsProcessor(LogitsProcessor):
+    """
+    :class:`transformers.LogitsProcessor` that enforces that when specified sequences appear, stop geration.
+
+    Args:
+        stop_words_ids (:obj:`List[List[int]]`):
+            List of list of token ids of stop ids. In order to get the tokens of the words
+            that should not appear in the generated text, use :obj:`tokenizer(bad_word,
+            add_prefix_space=True).input_ids`.
+        eos_token_id (:obj:`int`):
+            The id of the `end-of-sequence` token.
+    """
+
+    def __init__(self, stop_words_ids: Iterable[Iterable[int]], eos_token_id: int):
+
+        if not isinstance(stop_words_ids, List) or len(stop_words_ids) == 0:
+            raise ValueError(
+                f"`stop_words_ids` has to be a non-emtpy list, but is {stop_words_ids}."
+            )
+        if any(not isinstance(bad_word_ids, list) for bad_word_ids in stop_words_ids):
+            raise ValueError(
+                f"`stop_words_ids` has to be a list of lists, but is {stop_words_ids}."
+            )
+        if any(
+            any(
+                (not isinstance(token_id, (int, np.integer)) or token_id < 0)
+                for token_id in stop_word_ids
+            )
+            for stop_word_ids in stop_words_ids
+        ):
+            raise ValueError(
+                f"Each list in `stop_words_ids` has to be a list of positive integers, but is {stop_words_ids}."
+            )
+
+        self.stop_words_ids = list(
+            filter(
+                lambda bad_token_seq: bad_token_seq != [eos_token_id], stop_words_ids
+            )
+        )
+        self.eos_token_id = eos_token_id
+        for stop_token_seq in self.stop_words_ids:
+            assert (
+                len(stop_token_seq) > 0
+            ), "Stop words token sequences {} cannot have an empty list".format(
+                stop_words_ids
+            )
+
+    def __call__(
+        self, input_ids: torch.LongTensor, scores: torch.FloatTensor
+    ) -> torch.FloatTensor:
+        stopped_samples = self._calc_stopped_samples(input_ids)
+        for i, should_stop in enumerate(stopped_samples):
+            if should_stop:
+                scores[i, self.eos_token_id] = float(2**15)
+        return scores
+
+    def _tokens_match(self, prev_tokens: torch.LongTensor, tokens: List[int]) -> bool:
+        if len(tokens) == 0:
+            # if bad word tokens is just one token always ban it
+            return True
+        elif len(tokens) > len(prev_tokens):
+            # if bad word tokens are longer then prev input_ids they can't be equal
+            return False
+        elif prev_tokens[-len(tokens) :].tolist() == tokens:
+            # if tokens match
+            return True
+        else:
+            return False
+
+    def _calc_stopped_samples(self, prev_input_ids: Iterable[int]) -> Iterable[int]:
+        stopped_samples = []
+        for prev_input_ids_slice in prev_input_ids:
+            match = False
+            for stop_token_seq in self.stop_words_ids:
+                if self._tokens_match(prev_input_ids_slice, stop_token_seq):
+                    # if tokens do not match continue
+                    match = True
+                    break
+            stopped_samples.append(match)
+
+        return stopped_samples
+
+
+def top_k_logits(logits, top_k=0, top_p=0.0, filter_value=-float("Inf")):
+    """This function has been mostly taken from huggingface conversational
+    ai code at
+        https://medium.com/huggingface/how-to-build-a-state-of-the-art-
+             conversational-ai-with-transfer-learning-2d818ac26313"""
+
+    if top_k > 0:
+        # Remove all tokens with a probability less than the
+        # last token of the top-k
+        indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
+        logits[indices_to_remove] = filter_value
+
+    if top_p > 0.0:
+        # Cconvert to 1D
+        sorted_logits, sorted_indices = torch.sort(logits, descending=True, dim=-1)
+        cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
+
+        # Remove tokens with cumulative probability above the threshold
+        sorted_indices_to_remove = cumulative_probs > top_p
+        # Shift the indices to the right to keep also the first token
+        # above the threshold
+        sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
+        sorted_indices_to_remove[..., 0] = 0
+        for i in range(sorted_indices.size(0)):
+            indices_to_remove = sorted_indices[i][sorted_indices_to_remove[i]]
+            logits[i][indices_to_remove] = filter_value
+
+    return logits
+
+
+def switch(val1, val2, boolean):
+    boolean = boolean.type_as(val1)
+    return (1 - boolean) * val1 + boolean * val2

monkey_model/special_tokens_map.json

+{
+  "pad_token": "<|endoftext|>"
+}

monkey_model/tokenizer_config.json

+{
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_qwen.QWenTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 2048,
+  "padding_side": "right",
+  "tokenizer_class": "QWenTokenizer"
+}

monkey_model/visual.py

+# Copyright (c) Alibaba Cloud.
+#
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+from collections import OrderedDict
+import math
+import requests
+from io import BytesIO
+from functools import partial
+from PIL import Image
+from typing import Callable, Optional, Sequence, Tuple, List
+import numpy as np
+
+import torch
+from torch import nn
+from torch.nn import functional as F
+from torch.nn.init import trunc_normal_
+from torchvision import transforms
+from torchvision.transforms import InterpolationMode
+def reconstruct_matrix(windows):
+    temp =[]
+    for col in windows:
+        temp.append(torch.cat((col),dim=3))
+    all_img = torch.cat(temp,dim=2)
+    return all_img
+
+
+def sliding_window(matrix, window_size, stride):
+    b,c,height, width = matrix.shape
+    window_rows = (height - window_size[0]) // stride + 1
+    window_cols = (width - window_size[1]) // stride + 1
+    windows = []
+    for i in range(window_rows):
+        windows_col = []
+        for j in range(window_cols):
+            window = matrix[:,:, i*stride:i*stride+window_size[0],  j*stride:j*stride+window_size[1]]
+            windows_col.append(window)
+        windows.append(windows_col)
+    return windows
+
+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 Resampler(nn.Module):
+    """
+    A 2D perceiver-resampler network with one cross attention layers by
+        (grid_size**2) learnable queries and 2d sincos pos_emb
+    Outputs:
+        A tensor with the shape of (grid_size**2, embed_dim)
+    """
+    def __init__(
+            self,
+            grid_size,
+            embed_dim,
+            num_heads,
+            kv_dim=None,
+            norm_layer=nn.LayerNorm
+    ):
+        super().__init__()
+        self.num_queries = grid_size ** 2
+        self.embed_dim = embed_dim
+        self.num_heads = num_heads
+
+        self.pos_embed = nn.Parameter(
+            torch.from_numpy(get_2d_sincos_pos_embed(embed_dim, grid_size)).float()
+        ).requires_grad_(False)
+
+        self.query = nn.Parameter(torch.zeros(self.num_queries, embed_dim))
+        trunc_normal_(self.query, std=.02)
+
+        if kv_dim is not None and kv_dim != embed_dim:
+            self.kv_proj = nn.Linear(kv_dim, embed_dim, bias=False)
+        else:
+            self.kv_proj = nn.Identity()
+
+        self.attn = nn.MultiheadAttention(embed_dim, num_heads)
+        self.ln_q = norm_layer(embed_dim)
+        self.ln_kv = norm_layer(embed_dim)
+        
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+
+    def forward(self, x, attn_mask=None):
+
+        pos_embed = get_abs_pos(self.pos_embed, x.size(1))
+
+        x = self.kv_proj(x)
+        x = self.ln_kv(x).permute(1, 0, 2)
+
+        N = x.shape[1]
+        q = self.ln_q(self.query)
+        out = self.attn(
+            self._repeat(q, N) + self.pos_embed.unsqueeze(1),
+            x + pos_embed.unsqueeze(1),
+            x,
+            attn_mask=attn_mask)[0]
+        return out.permute(1, 0, 2)
+
+    def _repeat(self, query, N: int):
+        return query.unsqueeze(1).repeat(1, N, 1)
+
+
+
+class Lora_Adapter(nn.Module):
+    def __init__(self,
+                 d_model=None,
+                 out_feat=None,
+                 r=16,
+                 dropout=0.05):
+        super().__init__()
+        self.d_model = d_model
+        self.out_feat = out_feat
+        self.r = r
+
+        self.lora_scale = nn.Parameter(torch.ones(1))
+
+
+        self.lora_a = nn.Linear(self.d_model, self.r,bias=False)
+        self.lora_b = nn.Linear(self.r, self.out_feat,bias=False)
+
+        self.lora_dropout =  nn.Dropout(p=dropout)
+
+        with torch.no_grad():
+            nn.init.kaiming_uniform_(self.lora_a.weight, a=math.sqrt(5))
+            nn.init.zeros_(self.lora_b.weight)
+
+    def forward(self, x ):
+        #residual = x if residual is None else residual
+
+        x = self.lora_dropout(x)
+        down = self.lora_a(x)
+        up = self.lora_b(down)
+
+        up = up * self.lora_scale
+        output = up
+
+        return output
+
+
+class VisualAttention(nn.Module):
+    """self-attention layer class.
+
+    Self-attention layer takes input with size [s, b, h]
+    and returns output of the same size.
+    """
+
+    def __init__(self, embed_dim, num_heads,
+                 bias=True, kdim=None, vdim=None,lora_repeat_num=4):
+        super(VisualAttention, self).__init__()
+        self.embed_dim = embed_dim
+        self.kdim = kdim if kdim is not None else embed_dim
+        self.vdim = vdim if vdim is not None else embed_dim
+        self._qkv_same_embed_dim = self.kdim == embed_dim and self.vdim == embed_dim
+
+        self.num_heads = num_heads
+
+        # Per attention head and per partition values.
+        assert embed_dim % num_heads == 0
+        self.hidden_size_per_attention_head = embed_dim // num_heads
+        self.num_attention_heads_per_partition = num_heads
+        self.hidden_size_per_partition = embed_dim
+
+        # Strided linear layer.
+        assert self._qkv_same_embed_dim, 'Only Support SelfAttention Currently'
+        self.in_proj = nn.Linear(embed_dim, 3 * embed_dim)
+        self.in_proj_lora = []
+        for _ in range(lora_repeat_num):
+            self.in_proj_lora.append(Lora_Adapter(d_model=embed_dim,out_feat=3 * embed_dim))
+        self.in_proj_lora = nn.ModuleList(self.in_proj_lora)
+
+        self.out_proj = nn.Linear(embed_dim, embed_dim)
+        self.out_proj_lora = []
+        for _ in range(lora_repeat_num):
+            self.out_proj_lora.append(Lora_Adapter(d_model=embed_dim,out_feat=embed_dim))
+        self.out_proj_lora = nn.ModuleList(self.out_proj_lora)
+        self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
+
+    def forward(self, query, key, value, attn_mask = None,idx = None):
+        # query/key/value: [sq, b, h]
+        sq, b, _ = query.size()
+
+        assert query is key, 'Only Support Self-Attention Currently'
+        sk = sq
+        mixed_x_layer = self.in_proj(query)
+        if idx == None:
+            pass
+        else:
+            lora_res = self.in_proj_lora[idx](query)
+            mixed_x_layer += lora_res
+
+        # [sq, b, (np * 3 * hn)] --> [sq, b, np, 3 * hn]
+        new_tensor_shape = mixed_x_layer.size()[:-1] + \
+            (self.num_attention_heads_per_partition,
+             3 * self.hidden_size_per_attention_head)
+        mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
+
+        # [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
+        query_layer, key_layer, value_layer = mixed_x_layer.split(
+            self.hidden_size_per_attention_head, dim=-1)
+
+        # [sq, b, np, hn] -> [sq, b * np, hn]
+        query_layer = query_layer.view(sq,
+            b * self.num_attention_heads_per_partition,
+            self.hidden_size_per_attention_head).transpose(0, 1)
+        # [sk, b, np, hn] -> [sk, b * np, hn]
+        key_layer = key_layer.view(sk,
+            b * self.num_attention_heads_per_partition,
+            self.hidden_size_per_attention_head).transpose(0, 1)
+
+        q_scaled = query_layer / self.norm_factor
+        if attn_mask is not None:
+            attention_probs = torch.baddbmm(attn_mask, q_scaled, key_layer.transpose(-2, -1))
+        else:
+            attention_probs = torch.bmm(q_scaled, key_layer.transpose(-2, -1))
+        attention_probs = attention_probs.softmax(dim=-1)
+
+        value_layer = value_layer.view(sk,
+            b * self.num_attention_heads_per_partition,
+            self.hidden_size_per_attention_head).transpose(0, 1)
+
+        # matmul: [b * np, sq, hn]
+        context_layer = torch.bmm(attention_probs, value_layer)
+
+        # change view [b, np, sq, hn]
+        context_layer = context_layer.view(b,
+            self.num_attention_heads_per_partition,
+            sq, self.hidden_size_per_attention_head)
+
+        # [b, np, sq, hn] --> [sq, b, np, hn]
+        context_layer = context_layer.permute(2, 0, 1, 3).contiguous()
+
+        # [sq, b, np, hn] --> [sq, b, hp]
+        new_context_layer_shape = context_layer.size()[:-2] + \
+            (self.hidden_size_per_partition,)
+        context_layer = context_layer.view(*new_context_layer_shape)
+
+        output = self.out_proj(context_layer)
+        if idx == None:
+            pass
+        else:
+            lora_res = self.out_proj_lora[idx](context_layer)
+            output += lora_res
+
+        return output
+
+
+class VisualAttentionBlock(nn.Module):
+    def __init__(
+            self,
+            d_model: int,
+            n_head: int,
+            mlp_ratio: float = 4.0,
+            act_layer: Callable = nn.GELU,
+            norm_layer: Callable = nn.LayerNorm,
+            is_cross_attention: bool = False,
+            lora_repeat_num = 4,
+    ):
+        super().__init__()
+
+        self.ln_1 = norm_layer(d_model)
+        if is_cross_attention:
+            self.ln_1_kv = norm_layer(d_model)
+
+        self.ln_2 = norm_layer(d_model)
+        mlp_width = int(d_model * mlp_ratio)
+        self.attn = VisualAttention(d_model, n_head,lora_repeat_num = lora_repeat_num)
+        self.mlp = nn.Sequential(OrderedDict([
+            ("c_fc", nn.Linear(d_model, mlp_width)),
+            ("gelu", act_layer()),
+            ("c_proj", nn.Linear(mlp_width, d_model))
+        ]))
+        self.mlp_lora = []
+        for _ in range(lora_repeat_num):
+            self.mlp_lora.append(Lora_Adapter(d_model=d_model,out_feat=d_model,r=32))
+        self.mlp_lora = nn.ModuleList(self.mlp_lora)
+
+
+    def attention(
+            self,
+            q_x: torch.Tensor,
+            k_x: Optional[torch.Tensor] = None,
+            v_x: Optional[torch.Tensor] = None,
+            attn_mask: Optional[torch.Tensor] = None,
+            idx = None
+    ):
+        k_x = k_x if k_x is not None else q_x
+        v_x = v_x if v_x is not None else q_x
+
+        attn_mask = attn_mask.to(q_x.dtype) if attn_mask is not None else None
+        return self.attn(q_x, k_x, v_x, attn_mask=attn_mask,idx=idx)
+
+    def forward(
+            self,
+            q_x: torch.Tensor,
+            k_x: Optional[torch.Tensor] = None,
+            v_x: Optional[torch.Tensor] = None,
+            attn_mask: Optional[torch.Tensor] = None,
+            idx = None
+    ):
+        k_x = self.ln_1_kv(k_x) if hasattr(self, "ln_1_kv") and k_x is not None else None
+        v_x = self.ln_1_kv(v_x) if hasattr(self, "ln_1_kv") and v_x is not None else None
+
+        x = q_x + self.attention(q_x=self.ln_1(q_x), k_x=k_x, v_x=v_x, attn_mask=attn_mask,idx=idx)
+        residual = x 
+        x = x + self.mlp(self.ln_2(x))
+
+        
+        if idx == None:
+            pass
+        else:
+            x += self.mlp_lora[idx](residual)
+        return x
+
+
+class TransformerBlock(nn.Module):
+    def __init__(
+            self,
+            width: int,
+            layers: int,
+            heads: int,
+            mlp_ratio: float = 4.0,
+            act_layer: Callable = nn.GELU,
+            norm_layer: Callable = nn.LayerNorm,
+            lora_repeat_num=4
+    ):
+        super().__init__()
+        self.width = width
+        self.layers = layers
+
+        self.resblocks = nn.ModuleList([
+            VisualAttentionBlock(
+                width, heads, mlp_ratio, act_layer=act_layer, norm_layer=norm_layer,lora_repeat_num=lora_repeat_num)
+            for _ in range(layers)
+        ])
+
+    def get_cast_dtype(self) -> torch.dtype:
+        return self.resblocks[0].mlp.c_fc.weight.dtype
+
+    def get_cast_device(self) -> torch.device:
+        return self.resblocks[0].mlp.c_fc.weight.device
+
+    def forward(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None,idx=None):
+        for r in self.resblocks:
+            x = r(x, attn_mask=attn_mask,idx=idx)
+        return x
+
+
+class VisionTransformer(nn.Module):
+
+    def __init__(
+            self,
+            image_size: int,
+            patch_size: int,
+            width: int,
+            layers: int,
+            heads: int,
+            mlp_ratio: float,
+            n_queries: int = 256,
+            output_dim: int = 512,
+            lora_repeat_num: int = 4,
+            **kwargs
+    ):
+        super().__init__()
+        image_height, image_width = self.image_size = (image_size, image_size)
+        patch_height, patch_width = self.patch_size = (patch_size, patch_size)
+        self.grid_size = (image_height // patch_height, image_width // patch_width)
+        self.output_dim = output_dim
+
+        mean = (0.48145466, 0.4578275, 0.40821073)
+        std = (0.26862954, 0.26130258, 0.27577711)
+        self.image_transform = transforms.Compose([
+            transforms.Resize(
+                (image_size, image_size),
+                interpolation=InterpolationMode.BICUBIC
+            ),
+            transforms.ToTensor(),
+            transforms.Normalize(mean=mean, std=std),
+        ])
+
+        self.conv1 = nn.Conv2d(in_channels=3, out_channels=width, kernel_size=patch_size, stride=patch_size, bias=False)
+
+        # class embeddings and positional embeddings
+        scale = width ** -0.5
+        self.positional_embedding = nn.Parameter(scale * torch.randn(256, width))
+
+        norm_layer = partial(nn.LayerNorm, eps=1e-6)
+        act_layer = nn.GELU
+
+        self.ln_pre = norm_layer(width)
+        self.transformer = TransformerBlock(
+            width,
+            layers,
+            heads,
+            mlp_ratio,
+            act_layer=act_layer,
+            norm_layer=norm_layer,
+            lora_repeat_num=lora_repeat_num
+        )
+
+        
+        self.attn_pool = Resampler(
+            grid_size=int(math.sqrt(n_queries)),
+            embed_dim=output_dim,
+            num_heads=output_dim // 128,
+            kv_dim=width,
+            norm_layer=norm_layer,
+        )
+        self.ln_post = norm_layer(output_dim)
+        self.proj = nn.Parameter((output_dim** -0.5) * torch.randn(output_dim, output_dim))
+
+    def forward(self, x: torch.Tensor,idx=None):
+        x = x.to(
+            dtype=self.transformer.get_cast_dtype(),
+            device=self.transformer.get_cast_device(),
+        )
+        # to patches
+        x = self.conv1(x)  # shape = [*, width, grid, grid]
+        x = x.reshape(x.shape[0], x.shape[1], -1)  # shape = [*, width, grid ** 2]
+        x = x.permute(0, 2, 1)  # shape = [*, grid ** 2, width]
+
+        x = x + get_abs_pos(self.positional_embedding, x.size(1))
+
+        x = self.ln_pre(x)
+
+        x = x.permute(1, 0, 2)  # NLD -> LND
+        x = self.transformer(x,idx=idx)
+        x = x.permute(1, 0, 2)  # LND -> NLD
+
+        x = self.attn_pool(x)
+        x = self.ln_post(x)
+        x = x @ self.proj
+        return x
+
+    def encode(self, image_paths: List[str]):
+        images = []
+        for image_path in image_paths:
+            if image_path.startswith("http://") or image_path.startswith("https://"):
+                image = Image.open(requests.get(image_path, stream=True).raw)
+            else:
+                image = Image.open(image_path)
+            image = image.convert("RGB")
+            images.append(self.image_transform(image))
+        images = torch.stack(images, dim=0)
+        B,C,H,W = images.shape
+        windows = sliding_window(images,window_size=(448,448),stride=448)
+
+
+        images_448 = F.interpolate(images, size=(448,448), mode='bicubic')
+        return windows,images_448
+if __name__ == "__main__":
+    pass
+    visual = VisionTransformer(
+        image_size= 896,
+        patch_size= 14,
+        width=1664,
+        layers = 48,
+        heads= 16,
+        mlp_ratio =  4.9231,
+        output_dim= 4096)
+
+    img = torch.randn(1,3,896,896)
+
+
+    from peft import LoraConfig, get_peft_model, prepare_model_for_int8_training, TaskType
+
+    # Define LoRA Config
+    lora_config = LoraConfig(
+        r=16,
+        lora_alpha=32,
+        target_modules=["in_proj","out_proj","c_fc","c_proj"],
+        lora_dropout=0.05,
+        bias="none",    
+    )
+    # prepare int-8 model for training
+    model = visual
+
+    # add LoRA adaptor
+    model = get_peft_model(model, lora_config)
+    model.print_trainable_parameters()
+    print(model)
+    print(visual)
+