423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-01-09 23:50:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

[fix] fp8: add Transformer Engine backend support (#9705)

Browse Source 
Co-authored-by: Yaowei Zheng <hiyouga@buaa.edu.cn>
This commit is contained in:
Santosh Bhavani
2025-12-31 18:18:02 -08:00
committed by GitHub
parent 6fe6bd290b
commit 355d5c5e5a
6 changed files with 128 additions and 70 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
src/llamafactory/hparams/parser.py
View File

@@ -142,15 +142,6 @@ def _verify_model_args(
logger.warning_rank0("We should use slow tokenizer for the Yi models. Change `use_fast_tokenizer` to False.")
model_args.use_fast_tokenizer = False
# Validate advanced training features
if model_args.fp8 and model_args.quantization_bit is not None:
raise ValueError("FP8 training is not compatible with quantization. Please disable one of them.")
if model_args.fp8_enable_fsdp_float8_all_gather and not model_args.fp8:
logger.warning_rank0("fp8_enable_fsdp_float8_all_gather requires fp8=True. Setting fp8=True.")
model_args.fp8 = True
def _check_extra_dependencies(
model_args: "ModelArguments",
finetuning_args: "FinetuningArguments",
@@ -347,6 +338,9 @@ def get_train_args(args: dict[str, Any] | list[str] | None = None) -> _TRAIN_CLS
if training_args.deepspeed is not None and (finetuning_args.use_galore or finetuning_args.use_apollo):
raise ValueError("GaLore and APOLLO are incompatible with DeepSpeed yet.")
if training_args.fp8 and training_args.quantization_bit is not None:
raise ValueError("FP8 training is not compatible with quantization. Please disable one of them.")
if model_args.infer_backend != EngineName.HF:
raise ValueError("vLLM/SGLang backend is only available for API, CLI and Web.")
@@ -363,6 +357,10 @@ def get_train_args(args: dict[str, Any] | list[str] | None = None) -> _TRAIN_CLS
_verify_model_args(model_args, data_args, finetuning_args)
_check_extra_dependencies(model_args, finetuning_args, training_args)
if training_args.fp8_enable_fsdp_float8_all_gather and not training_args.fp8:
logger.warning_rank0("fp8_enable_fsdp_float8_all_gather requires fp8=True. Setting fp8=True.")
model_args.fp8 = True
if (
training_args.do_train
and finetuning_args.finetuning_type == "lora"