olmo.py

# SPDX-License-Identifier: Apache-2.0

# Adapted from
# https://github.com/huggingface/transformers/blob/v4.40.1/src/transformers/models/olmo/modeling_olmo.py
# Copyright 2024 The vLLM team.
# Copyright 2024 EleutherAI and the HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
#
# 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.
"""Inference-only OLMo model compatible with HuggingFace weights."""
from collections.abc import Iterable
from typing import Optional, Union

import torch
from torch import nn
from transformers import OlmoConfig

from vllm.attention import Attention
from vllm.compilation.decorators import support_torch_compile
from vllm.config import CacheConfig, VllmConfig
from vllm.distributed import get_pp_group, get_tensor_model_parallel_world_size
from vllm.model_executor.layers.activation import SiluAndMul
from vllm.model_executor.layers.linear import (MergedColumnParallelLinear,
                                               QKVParallelLinear,
                                               RowParallelLinear)
from vllm.model_executor.layers.logits_processor import LogitsProcessor
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.rotary_embedding import get_rope
from vllm.model_executor.layers.vocab_parallel_embedding import (
    ParallelLMHead, VocabParallelEmbedding)
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.sequence import IntermediateTensors

from .interfaces import SupportsPP
from .utils import (is_pp_missing_parameter,
                    make_empty_intermediate_tensors_factory, make_layers,
                    maybe_prefix)


class OlmoAttention(nn.Module):
    """
    This is the attention block where the output is computed as
    ``Attention(LN(x))`` in ``MLP(LN(x + Attention(LN(x))))``
    (plus another skip connection).
    """

    def __init__(
        self,
        config: OlmoConfig,
        cache_config: Optional[CacheConfig] = None,
        quant_config: Optional[QuantizationConfig] = None,
        prefix: str = "",
    ):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        tensor_model_parallel_world_size = (
            get_tensor_model_parallel_world_size())
        self.total_num_heads = config.num_attention_heads

        assert self.hidden_size % self.total_num_heads == 0
        assert self.total_num_heads % tensor_model_parallel_world_size == 0

        self.num_heads = (self.total_num_heads //
                          tensor_model_parallel_world_size)
        self.head_dim = self.hidden_size // self.total_num_heads
        self.max_position_embeddings = config.max_position_embeddings
        self.rope_theta = config.rope_theta
        self.clip_qkv = config.clip_qkv

        # Attention input projection. Projects x -> (q, k, v)
        self.qkv_proj = QKVParallelLinear(
            self.hidden_size,
            self.head_dim,
            self.total_num_heads,
            bias=config.attention_bias,
            quant_config=quant_config,
        )

        # Rotary embeddings.
        self.rotary_emb = get_rope(
            self.head_dim,
            rotary_dim=self.head_dim,
            max_position=self.max_position_embeddings,
            base=self.rope_theta,
        )
        self.scaling = self.head_dim**-0.5
        self.attn = Attention(self.num_heads,
                              self.head_dim,
                              scale=self.scaling,
                              cache_config=cache_config,
                              quant_config=quant_config,
                              prefix=f"{prefix}.attn")

        # Attention output projection.
        self.o_proj = RowParallelLinear(
            self.hidden_size,
            self.hidden_size,
            bias=config.attention_bias,
            quant_config=quant_config,
        )

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
    ) -> torch.Tensor:
        qkv, _ = self.qkv_proj(hidden_states)
        if self.clip_qkv is not None:
            qkv.clamp_(min=-self.clip_qkv, max=self.clip_qkv)
        q, k, v = qkv.chunk(chunks=3, dim=-1)
        q, k = self.rotary_emb(positions, q, k)
        attn_output = self.attn(q, k, v)
        output, _ = self.o_proj(attn_output)
        return output


class OlmoMLP(nn.Module):
    """
    This is the MLP block where the output is computed as
    ``MLP(LN(x))`` in ``MLP(LN(x + Attention(LN(x))))``
    (plus another skip connection).
    """

    def __init__(
        self,
        config: OlmoConfig,
        quant_config: Optional[QuantizationConfig] = None,
    ):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.intermediate_size

        # Feed-forward input projection.
        self.gate_up_proj = MergedColumnParallelLinear(
            self.hidden_size,
            [self.intermediate_size] * 2,
            bias=False,
            quant_config=quant_config,
        )

        # Activation function.
        self.act_fn = SiluAndMul()

        # Feed-forward output projection.
        self.down_proj = RowParallelLinear(
            self.intermediate_size,
            self.hidden_size,
            bias=False,
            quant_config=quant_config,
        )

    def forward(
        self,
        x: torch.Tensor,
    ) -> torch.Tensor:
        gate_up, _ = self.gate_up_proj(x)
        x = self.act_fn(gate_up)
        x, _ = self.down_proj(x)
        return x


class OlmoDecoderLayer(nn.Module):
    """
    This is a typical transformer block where the output is
    computed as ``MLP(LN(x + Attention(LN(x))))``
    (plus another skip connection).
    """

    def __init__(self,
                 config: OlmoConfig,
                 cache_config: Optional[CacheConfig] = None,
                 quant_config: Optional[QuantizationConfig] = None,
                 prefix: str = ""):
        super().__init__()
        # Attention block.
        self.self_attn = OlmoAttention(config,
                                       cache_config,
                                       quant_config,
                                       prefix=f"{prefix}.self_attn")

        # MLP block.
        self.mlp = OlmoMLP(config, quant_config)

        # LayerNorm
        self.input_layernorm = nn.LayerNorm(config.hidden_size,
                                            elementwise_affine=False,
                                            bias=False)
        self.post_attention_layernorm = nn.LayerNorm(config.hidden_size,
                                                     elementwise_affine=False,
                                                     bias=False)

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
    ) -> tuple[torch.Tensor, Optional[tuple[torch.Tensor, torch.Tensor]]]:
        # Attention block.
        residual = hidden_states
        hidden_states = self.input_layernorm(hidden_states)
        hidden_states = self.self_attn(positions, hidden_states)
        hidden_states = hidden_states + residual

        # MLP block.
        residual = hidden_states
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = residual + hidden_states
        return hidden_states


@support_torch_compile
class OlmoModel(nn.Module):

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()

        config = vllm_config.model_config.hf_config
        cache_config = vllm_config.cache_config
        quant_config = vllm_config.quant_config

        self.config = config

        self.embed_tokens = VocabParallelEmbedding(config.vocab_size,
                                                   config.hidden_size)
        self.start_layer, self.end_layer, self.layers = make_layers(
            config.num_hidden_layers,
            lambda prefix: OlmoDecoderLayer(
                config, cache_config, quant_config, prefix=prefix),
            prefix=f"{prefix}.layers")
        self.norm = nn.LayerNorm(config.hidden_size,
                                 elementwise_affine=False,
                                 bias=False)
        self.make_empty_intermediate_tensors = (
            make_empty_intermediate_tensors_factory(["hidden_states"],
                                                    config.hidden_size))

    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.embed_tokens(input_ids)

    def forward(
        self,
        input_ids: torch.Tensor,
        positions: torch.Tensor,
        intermediate_tensors: Optional[IntermediateTensors],
        inputs_embeds: Optional[torch.Tensor] = None,
    ) -> Union[torch.Tensor, IntermediateTensors]:
        """
        :param input_ids: A tensor of shape `(batch_size, seq_len)`.
        """
        if get_pp_group().is_first_rank:
            if inputs_embeds is not None:
                hidden_states = inputs_embeds
            else:
                hidden_states = self.get_input_embeddings(input_ids)
        else:
            assert intermediate_tensors is not None
            hidden_states = intermediate_tensors["hidden_states"]

        # Apply blocks one-by-one.
        for layer in self.layers[self.start_layer:self.end_layer]:
            # shape: (batch_size, seq_len, d_model)
            hidden_states = layer(positions, hidden_states)

        if not get_pp_group().is_last_rank:
            return IntermediateTensors({"hidden_states": hidden_states})
        # Apply final layer norm.
        # shape: (batch_size, seq_len or 1, d_model)
        hidden_states = self.norm(hidden_states)
        return hidden_states


class OlmoForCausalLM(nn.Module, SupportsPP):
    """
    Extremely barebones HF model wrapper.
    """

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()
        config = vllm_config.model_config.hf_config
        quant_config = vllm_config.quant_config
        self.config = config
        self.model = OlmoModel(vllm_config=vllm_config,
                               prefix=maybe_prefix(prefix, "model"))
        if config.tie_word_embeddings:
            self.lm_head = self.model.embed_tokens
        else:
            self.unpadded_vocab_size = config.vocab_size
            self.lm_head = ParallelLMHead(
                self.unpadded_vocab_size,
                config.hidden_size,
                org_num_embeddings=config.vocab_size,
                quant_config=quant_config,
            )
        self.logits_processor = LogitsProcessor(config.vocab_size)
        self.make_empty_intermediate_tensors = (
            self.model.make_empty_intermediate_tensors)

    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.model.get_input_embeddings(input_ids)

    def forward(
        self,
        input_ids: torch.Tensor,
        positions: torch.Tensor,
        intermediate_tensors: Optional[IntermediateTensors] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
    ) -> Union[torch.Tensor, IntermediateTensors]:
        hidden_states = self.model(
            input_ids=input_ids,
            positions=positions,
            intermediate_tensors=intermediate_tensors,
            inputs_embeds=inputs_embeds,
        )
        return hidden_states

    def compute_logits(
        self,
        hidden_states: torch.Tensor,
        sampling_metadata: SamplingMetadata,
    ) -> Optional[torch.Tensor]:
        logits = self.logits_processor(self.lm_head, hidden_states,
                                       sampling_metadata)
        return logits

    def load_weights(self, weights: Iterable[tuple[str,
                                                   torch.Tensor]]) -> set[str]:
        stacked_params_mapping = [
            # (param_name, shard_name, shard_id)
            ("qkv_proj", "q_proj", "q"),
            ("qkv_proj", "k_proj", "k"),
            ("qkv_proj", "v_proj", "v"),
            ("gate_up_proj", "gate_proj", 0),
            ("gate_up_proj", "up_proj", 1),
        ]
        params_dict = dict(self.named_parameters(remove_duplicate=False))
        loaded_params: set[str] = set()
        for name, loaded_weight in weights:
            if "rotary_emb.inv_freq" in name:
                continue
            if ("rotary_emb.cos_cached" in name
                    or "rotary_emb.sin_cached" in name):
                # Models trained using ColossalAI may include these tensors in
                # the checkpoint. Skip them.
                continue
            # With tie_word_embeddings, we can skip lm_head.weight
            # The weight might appear unnecessarily in the files if the model is
            # processed with quantization, LoRA, fine-tuning, etc.
            if self.config.tie_word_embeddings and "lm_head.weight" in name:
                continue
            for (param_name, weight_name, shard_id) in stacked_params_mapping:
                if weight_name not in name:
                    continue
                name = name.replace(weight_name, param_name)
                # Skip loading extra bias for GPTQ models.
                if name.endswith(".bias") and name not in params_dict:
                    continue
                if is_pp_missing_parameter(name, self):
                    continue
                param = params_dict[name]
                weight_loader = param.weight_loader
                weight_loader(param, loaded_weight, shard_id)
                break
            else:
                # Skip loading extra bias for GPTQ models.
                if name.endswith(".bias") and name not in params_dict:
                    continue
                if is_pp_missing_parameter(name, self):
                    continue
                param = params_dict[name]
                weight_loader = getattr(param, "weight_loader",
                                        default_weight_loader)
                weight_loader(param, loaded_weight)
            loaded_params.add(name)
        return loaded_params