Files
LLaMA-Factory/src/llamafactory/model/patcher.py

469 lines
20 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Copyright 2025 the LlamaFactory team.
#
# 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 types import MethodType
from typing import TYPE_CHECKING, Any
import torch
from peft import PeftModel
from transformers import GenerationMixin, PreTrainedModel, PreTrainedTokenizerBase
from transformers.integrations import is_deepspeed_zero3_enabled
from transformers.modeling_utils import is_fsdp_enabled
from ..extras import logging
from ..extras.misc import infer_optim_dtype
from ..extras.packages import is_transformers_version_greater_than
from .model_utils.attention import configure_attn_implementation, print_attn_implementation
from .model_utils.checkpointing import prepare_model_for_training
from .model_utils.embedding import resize_embedding_layer
from .model_utils.kv_cache import configure_kv_cache
from .model_utils.longlora import configure_longlora
from .model_utils.moe import add_z3_leaf_module, configure_moe
from .model_utils.quantization import configure_quantization
from .model_utils.rope import configure_rope
from .model_utils.valuehead import prepare_valuehead_model
from .model_utils.visual import autocast_projector_dtype, configure_visual_model
if TYPE_CHECKING:
from transformers import PretrainedConfig, PreTrainedTokenizer, ProcessorMixin
from trl import AutoModelForCausalLMWithValueHead
from ..hparams import ModelArguments
if is_transformers_version_greater_than("4.57.0"):
from transformers.models.qwen3_omni_moe import modeling_qwen3_omni_moe
logger = logging.get_logger(__name__)
def patch_qwen3_omni_moe_thinker_text_sparse_moe_block():
if is_transformers_version_greater_than("4.57.0") and not is_transformers_version_greater_than("4.58.0"):
from .model_utils.moe import Qwen3OmniMoeThinkerTextSparseMoeBlock
logger.warning_rank0(
"You are using transformers with 4.x version, the Qwen3OmniMoeThinkerTextSparseMoeBlock will have some issues about deepspeed zero2 and fsdp2 training, so that we patched this model to avoid it. Transformers v5.0.0rc0 has fixed the issue, you can also try to update the transformers to using qwen3_omni. See more information on https://github.com/hiyouga/LLaMA-Factory/issues/9628."
)
modeling_qwen3_omni_moe.Qwen3OmniMoeThinkerTextSparseMoeBlock = Qwen3OmniMoeThinkerTextSparseMoeBlock
def _check_fla_dependencies() -> None:
"""Check that the FLA dependencies required for varlen GDN forwarding are available.
Requires ``flash-linear-attention >= 0.4.1`` (which exposes the varlen
``causal_conv1d`` under ``fla.modules.convolution`` and the
``chunk_gated_delta_rule`` / ``fused_recurrent_gated_delta_rule`` kernels
under ``fla.ops.gated_delta_rule``). Raises ``ImportError`` with an
actionable message otherwise.
"""
try:
from fla.modules.convolution import causal_conv1d # noqa: F401
from fla.ops.gated_delta_rule import ( # noqa: F401
chunk_gated_delta_rule,
fused_recurrent_gated_delta_rule,
)
except ImportError as exc:
raise ImportError(
"Qwen3.5 packing-seq forwarding requires `flash-linear-attention>=0.4.1` "
"(provides `fla.modules.convolution.causal_conv1d` and "
"`fla.ops.gated_delta_rule.{chunk,fused_recurrent}_gated_delta_rule`). "
"Please install/upgrade it."
) from exc
def patch_qwen3_5_forward(model: "PreTrainedModel") -> None:
"""Patch the forward method of Qwen3_5ForConditionalGeneration to support cu_seqlens input only patch when do training.
Refer to: https://github.com/axolotl-ai-cloud/axolotl/blob/main/src/axolotl/monkeypatch/models/qwen3_5/modeling.py.
"""
if is_transformers_version_greater_than("5.2.0"):
from transformers.models.qwen3_5.modeling_qwen3_5 import apply_mask_to_padding_states
from torch.nn import functional as F
from transformers.modeling_flash_attention_utils import prepare_fa_kwargs_from_position_ids
_check_fla_dependencies()
from fla.modules.convolution import causal_conv1d as fla_causal_conv1d
from fla.ops.gated_delta_rule import chunk_gated_delta_rule
def _patched_decoder_forward(
self,
hidden_states: torch.Tensor,
position_embeddings: tuple[torch.Tensor, torch.Tensor],
attention_mask: torch.Tensor | None = None,
position_ids: torch.LongTensor | None = None,
past_key_values=None,
cache_position: torch.LongTensor | None = None,
**kwargs,
) -> torch.FloatTensor:
"""Decoder layer forward that passes position_ids through to linear attention."""
residual = hidden_states
hidden_states = self.input_layernorm(hidden_states)
if self.layer_type == "linear_attention":
hidden_states = self.linear_attn(
hidden_states=hidden_states,
cache_params=past_key_values,
cache_position=cache_position,
attention_mask=attention_mask,
position_ids=position_ids, # passing position_ids to linear attention
)
elif self.layer_type == "full_attention":
hidden_states, _ = self.self_attn(
hidden_states=hidden_states,
attention_mask=attention_mask,
position_ids=position_ids,
past_key_values=past_key_values,
cache_position=cache_position,
position_embeddings=position_embeddings,
**kwargs,
)
hidden_states = residual + hidden_states
residual = hidden_states
hidden_states = self.post_attention_layernorm(hidden_states)
hidden_states = self.mlp(hidden_states)
if isinstance(hidden_states, tuple): # MoE returns (hidden_states, router_logits)
hidden_states, _ = hidden_states
hidden_states = residual + hidden_states
return hidden_states
# gdn forward (training only, cache_params is always None)
def _patch_gdn_forward(
self,
hidden_states: torch.Tensor,
cache_params=None,
cache_position: torch.LongTensor | None = None,
attention_mask: torch.Tensor | None = None,
position_ids: torch.LongTensor | None = None,
):
# @kuangdd fix: here attention_mask is None
hidden_states = apply_mask_to_padding_states(hidden_states, attention_mask)
batch_size, seq_len, _ = hidden_states.shape
# Qwen3.5 VL passes 3-D MRoPE position_ids ([axes, B, T]); collapse to [B, T].
if position_ids is not None and position_ids.ndim == 3:
position_ids = position_ids[0]
# `prepare_fa_kwargs_from_position_ids` would crash on None; guard for safety.
cu_seqlens = (
prepare_fa_kwargs_from_position_ids(position_ids)[0][0]
if position_ids is not None
else None
)
# FLA varlen kernels expect [B, T, D] layout, not [B, D, T] like the
# standard causal-conv1d path that the upstream forward uses.
mixed_qkv = self.in_proj_qkv(hidden_states)
z = self.in_proj_z(hidden_states)
z = z.reshape(batch_size, seq_len, -1, self.head_v_dim)
b = self.in_proj_b(hidden_states)
a = self.in_proj_a(hidden_states)
# FLA's causal_conv1d returns (out, final_state); we don't use the state here.
mixed_qkv, _ = fla_causal_conv1d(
x=mixed_qkv,
weight=self.conv1d.weight.squeeze(1),
bias=self.conv1d.bias,
activation=self.activation,
cu_seqlens=cu_seqlens,
)
query, key, value = torch.split(
mixed_qkv,
[
self.key_dim,
self.key_dim,
self.value_dim,
],
dim=-1,
)
query = query.reshape(batch_size, seq_len, -1, self.head_k_dim)
key = key.reshape(batch_size, seq_len, -1, self.head_k_dim)
value = value.reshape(batch_size, seq_len, -1, self.head_v_dim)
beta = b.sigmoid()
# If the model is loaded in fp16, without the .float() here, A might be -inf
g = -self.A_log.float().exp() * F.softplus(a.float() + self.dt_bias)
if self.num_v_heads // self.num_k_heads > 1:
query = query.repeat_interleave(self.num_v_heads // self.num_k_heads, dim=2)
key = key.repeat_interleave(self.num_v_heads // self.num_k_heads, dim=2)
core_attn_out, _ = chunk_gated_delta_rule(
query,
key,
value,
g=g,
beta=beta,
initial_state=None,
output_final_state=False,
use_qk_l2norm_in_kernel=True,
**({"cu_seqlens": cu_seqlens} if cu_seqlens is not None else {}),
)
core_attn_out = core_attn_out.reshape(-1, self.head_v_dim)
z = z.reshape(-1, self.head_v_dim)
core_attn_out = self.norm(core_attn_out, z)
core_attn_out = core_attn_out.reshape(batch_size, seq_len, -1)
output = self.out_proj(core_attn_out)
return output
if model.config.architectures[0] == "Qwen3_5ForConditionalGeneration":
from transformers.models.qwen3_5.modeling_qwen3_5 import Qwen3_5DecoderLayer, Qwen3_5GatedDeltaNet
Qwen3_5DecoderLayer.forward = _patched_decoder_forward
Qwen3_5GatedDeltaNet.forward = _patch_gdn_forward
elif model.config.architectures[0] == "Qwen3_5MoeForConditionalGeneration":
from transformers.models.qwen3_5_moe.modeling_qwen3_5_moe import (
Qwen3_5MoeDecoderLayer,
Qwen3_5MoeGatedDeltaNet,
)
Qwen3_5MoeDecoderLayer.forward = _patched_decoder_forward
Qwen3_5MoeGatedDeltaNet.forward = _patch_gdn_forward
logger.info_rank0("Patched Qwen3.5 decoder forward to support cu_seqlens input only patch when do training.")
def patch_youtu_vl_model(model: "PreTrainedModel") -> None:
original_forward = model.forward
def forward(self, *args, **kwargs):
outputs = original_forward(*args, **kwargs)
if "loss" not in outputs and "labels" in kwargs:
logits = outputs.get("logits")
labels = kwargs.get("labels")
if logits is not None and labels is not None:
shift_logits = logits[..., :-1, :].contiguous()
shift_labels = labels[..., 1:].contiguous()
loss_fct = torch.nn.CrossEntropyLoss()
loss = loss_fct(shift_logits.view(-1, self.config.vocab_size), shift_labels.view(-1))
outputs["loss"] = loss
return outputs
model.forward = MethodType(forward, model)
def patch_tokenizer(tokenizer: "PreTrainedTokenizer", model_args: "ModelArguments") -> None:
if "PreTrainedTokenizerBase" not in str(tokenizer._pad.__func__):
tokenizer._pad = MethodType(PreTrainedTokenizerBase._pad, tokenizer)
if model_args.model_max_length is not None and tokenizer.model_max_length < model_args.model_max_length:
tokenizer.model_max_length = model_args.model_max_length # enlarge the tokenizer max length
if model_args.add_tokens is not None:
num_added_tokens = tokenizer.add_tokens(new_tokens=model_args.add_tokens, special_tokens=False)
logger.info_rank0("Add tokens {} to tokenizer's vocabulary.".format(",".join(model_args.add_tokens)))
if num_added_tokens > 0 and not model_args.resize_vocab:
model_args.resize_vocab = True
logger.warning_rank0("New tokens have been added, changed `resize_vocab` to True.")
if model_args.add_special_tokens is not None:
num_added_special_tokens = tokenizer.add_tokens(new_tokens=model_args.add_special_tokens, special_tokens=True)
logger.info_rank0(
"Add special tokens {} to tokenizer's vocabulary.".format(",".join(model_args.add_special_tokens))
)
if num_added_special_tokens > 0 and not model_args.resize_vocab:
model_args.resize_vocab = True
logger.warning_rank0("New special tokens have been added, changed `resize_vocab` to True.")
def patch_processor(
processor: "ProcessorMixin",
tokenizer: "PreTrainedTokenizer",
model_args: "ModelArguments",
) -> None:
setattr(processor, "tokenizer", tokenizer)
setattr(processor, "image_max_pixels", model_args.image_max_pixels)
setattr(processor, "image_min_pixels", model_args.image_min_pixels)
setattr(processor, "image_do_pan_and_scan", model_args.image_do_pan_and_scan)
setattr(processor, "crop_to_patches", model_args.crop_to_patches)
setattr(processor, "video_max_pixels", model_args.video_max_pixels)
setattr(processor, "video_min_pixels", model_args.video_min_pixels)
setattr(processor, "video_fps", model_args.video_fps)
setattr(processor, "video_maxlen", model_args.video_maxlen)
setattr(processor, "use_audio_in_video", model_args.use_audio_in_video)
setattr(processor, "audio_sampling_rate", model_args.audio_sampling_rate)
def patch_config(
config: "PretrainedConfig",
tokenizer: "PreTrainedTokenizer",
model_args: "ModelArguments",
init_kwargs: dict[str, Any],
is_trainable: bool,
) -> None:
if model_args.compute_dtype is None: # priority: bf16 > fp16 > fp32
if model_args.infer_dtype != "auto" and not is_trainable:
model_args.compute_dtype = getattr(torch, model_args.infer_dtype)
else:
model_args.compute_dtype = infer_optim_dtype(model_dtype=getattr(config, "torch_dtype", None))
configure_attn_implementation(config, model_args)
configure_rope(config, model_args)
configure_longlora(config, model_args, is_trainable)
configure_quantization(config, tokenizer, model_args, is_trainable, init_kwargs)
configure_moe(config, model_args, is_trainable)
configure_visual_model(config)
configure_kv_cache(config, model_args, is_trainable)
if getattr(config, "model_type", None) == "qwen":
setattr(config, "use_flash_attn", model_args.flash_attn == "fa2")
for dtype_name, dtype in [("fp16", torch.float16), ("bf16", torch.bfloat16), ("fp32", torch.float32)]:
setattr(config, dtype_name, model_args.compute_dtype == dtype)
if getattr(config, "model_type", None) == "minicpmo":
setattr(config, "init_audio", True)
setattr(config, "init_tts", False)
# replace the top-k gating method
if getattr(config, "model_type", None) == "kimi_vl" and is_trainable:
setattr(config.text_config, "topk_method", "greedy")
architectures = getattr(config, "architectures", None)
if isinstance(architectures, list) and "InternVLChatModel" in architectures:
raise ValueError(
"Please download the internvl models in a Hugging Facecompatible format "
"(for example, https://huggingface.co/OpenGVLab/InternVL3-8B-hf)."
)
if isinstance(architectures, list) and "LlavaLlamaForCausalLM" in architectures:
raise ValueError("Please download llava models with hf-compatible format: https://huggingface.co/llava-hf")
if getattr(config, "model_type", None) == "internlm3" and not is_transformers_version_greater_than("4.47.1"):
raise RuntimeError("InternLM3 model requires transformers>=4.47.1, please upgrade it.")
if getattr(config, "model_type", None) == "lfm2_vl" and not is_transformers_version_greater_than("4.58.0"):
raise RuntimeError(
"LFM2.5-VL model requires transformers>=4.58.0 or install from commit: "
"pip install git+https://github.com/huggingface/transformers.git@3c2517727ce28a30f5044e01663ee204deb1cdbe"
)
if getattr(config, "model_type", None) == "qwen3_omni_moe":
patch_qwen3_omni_moe_thinker_text_sparse_moe_block()
# deepspeed zero3 is not compatible with low_cpu_mem_usage
init_kwargs["low_cpu_mem_usage"] = model_args.low_cpu_mem_usage and (not is_deepspeed_zero3_enabled())
# fsdp/deepspeed zero3 does not need device map
if not (is_deepspeed_zero3_enabled() or is_fsdp_enabled()) and init_kwargs["low_cpu_mem_usage"]:
if "device_map" not in init_kwargs and model_args.device_map:
init_kwargs["device_map"] = model_args.device_map # device map requires low_cpu_mem_usage=True
if init_kwargs.get("device_map", None) == "auto":
init_kwargs["offload_folder"] = model_args.offload_folder
def patch_model(
model: "PreTrainedModel",
tokenizer: "PreTrainedTokenizer",
model_args: "ModelArguments",
is_trainable: bool,
add_valuehead: bool,
) -> None:
gen_config = model.generation_config # check and fix generation config
if not gen_config.do_sample and (
(gen_config.temperature is not None and gen_config.temperature != 1.0)
or (gen_config.top_p is not None and gen_config.top_p != 1.0)
or (gen_config.typical_p is not None and gen_config.typical_p != 1.0)
):
gen_config.do_sample = True
if getattr(model.config, "model_type", None) not in ["minicpmv", "minicpmo"] and "GenerationMixin" not in str(
model.generate.__func__
):
model.generate = MethodType(GenerationMixin.generate, model)
if add_valuehead:
prepare_valuehead_model(model)
if model_args.resize_vocab:
resize_embedding_layer(
model,
tokenizer,
new_special_tokens_config=getattr(model_args, "_special_token_descriptions", None),
init_special_tokens=model_args.init_special_tokens,
)
if is_trainable:
if getattr(model.config, "model_type", None) == "gemma3n":
setattr(model_args, "disable_gradient_checkpointing", True)
if getattr(model.config, "model_type", None) == "youtu_vl":
patch_youtu_vl_model(model)
prepare_model_for_training(model, model_args)
autocast_projector_dtype(model, model_args)
add_z3_leaf_module(model)
if getattr(model.config, "model_type", None) in ["qwen3_5", "qwen3_5_moe"] and model_args.flash_attn == "fa2":
patch_qwen3_5_forward(model)
if not model_args.use_unsloth:
print_attn_implementation(model.config)
try:
model.add_model_tags(["llama-factory"])
except Exception:
logger.warning_rank0("Cannot properly tag the model.")
def patch_valuehead_model(model: "AutoModelForCausalLMWithValueHead") -> None:
def tie_weights(self: "AutoModelForCausalLMWithValueHead") -> None:
if isinstance(self.pretrained_model, PreTrainedModel):
self.pretrained_model.tie_weights()
def get_input_embeddings(self: "AutoModelForCausalLMWithValueHead") -> torch.nn.Module:
if isinstance(self.pretrained_model, PreTrainedModel):
return self.pretrained_model.get_input_embeddings()
def get_output_embeddings(self: "AutoModelForCausalLMWithValueHead") -> torch.nn.Module:
if isinstance(self.pretrained_model, PreTrainedModel):
return self.pretrained_model.get_output_embeddings()
def create_or_update_model_card(self: "AutoModelForCausalLMWithValueHead", output_dir: str) -> None:
if isinstance(self.pretrained_model, PeftModel):
self.pretrained_model.create_or_update_model_card(output_dir)
def get_rope_index_func(self: "AutoModelForCausalLMWithValueHead"):
if isinstance(self.pretrained_model, PeftModel):
base_model = self.pretrained_model.base_model.model
else:
base_model = self.pretrained_model
if base_model and hasattr(base_model, "get_rope_index"):
return base_model.get_rope_index
elif base_model and hasattr(base_model, "model") and hasattr(base_model.model, "get_rope_index"):
return base_model.model.get_rope_index
else:
return None
ignore_modules = [name for name, _ in model.named_parameters() if "pretrained_model" in name]
setattr(model, "_keys_to_ignore_on_save", ignore_modules)
setattr(model, "tie_weights", MethodType(tie_weights, model))
setattr(model, "get_input_embeddings", MethodType(get_input_embeddings, model))
setattr(model, "get_output_embeddings", MethodType(get_output_embeddings, model))
setattr(model, "get_rope_index", get_rope_index_func(model))
setattr(model, "create_or_update_model_card", MethodType(create_or_update_model_card, model))