423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-02 11:42:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

support infer 4bit model on GPUs #3023

Browse Source 
Former-commit-id: eb259cc5738dfb383e4cc5d32579501c580e11b1
This commit is contained in:
hiyouga 2024-04-01 17:34:04 +08:00
parent cefe7f7bcf
commit b7468ea0a8
2 changed files with 14 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/llmtuner/hparams/model_args.py
View File

@ -53,6 +53,10 @@ class ModelArguments:
default=True, default=True,
metadata={"help": "Whether or not to use double quantization in int4 training."}, metadata={"help": "Whether or not to use double quantization in int4 training."},
) )
quantization_device_map: Optional[Literal["auto"]] = field(
default=None,
metadata={"help": "Device map used for loading the 4-bit quantized model, needs bitsandbytes>=0.43.0."},
)
rope_scaling: Optional[Literal["linear", "dynamic"]] = field( rope_scaling: Optional[Literal["linear", "dynamic"]] = field(
default=None, default=None,
metadata={"help": "Which scaling strategy should be adopted for the RoPE embeddings."}, metadata={"help": "Which scaling strategy should be adopted for the RoPE embeddings."},

16
src/llmtuner/model/patcher.py
View File

@ -208,11 +208,6 @@ def _configure_quantization(
logger.info("Quantizing model to {} bit.".format(model_args.export_quantization_bit)) logger.info("Quantizing model to {} bit.".format(model_args.export_quantization_bit))
elif model_args.quantization_bit is not None: # bnb elif model_args.quantization_bit is not None: # bnb
if is_deepspeed_zero3_enabled():
require_version("transformers>=4.39.0", "To fix: pip install transformers>=4.39.0")
require_version("accelerate>=0.28.0", "To fix: pip install accelerate>=0.28.0")
require_version("bitsandbytes>=0.43.0", "To fix: pip install bitsandbytes>=0.43.0")
if model_args.quantization_bit == 8: if model_args.quantization_bit == 8:
require_version("bitsandbytes>=0.37.0", "To fix: pip install bitsandbytes>=0.37.0") require_version("bitsandbytes>=0.37.0", "To fix: pip install bitsandbytes>=0.37.0")
init_kwargs["quantization_config"] = BitsAndBytesConfig(load_in_8bit=True) init_kwargs["quantization_config"] = BitsAndBytesConfig(load_in_8bit=True)
@ -227,7 +222,16 @@ def _configure_quantization(
bnb_4bit_quant_storage=model_args.compute_dtype, # crucial for fsdp qlora bnb_4bit_quant_storage=model_args.compute_dtype, # crucial for fsdp qlora
) )
init_kwargs["device_map"] = {"": get_current_device()} if is_deepspeed_zero3_enabled() or model_args.quantization_device_map == "auto":
if model_args.quantization_bit != 4:
raise ValueError("Only 4-bit quantized model can use auto device map.")
require_version("transformers>=4.39.0", "To fix: pip install transformers>=4.39.0")
require_version("accelerate>=0.28.0", "To fix: pip install accelerate>=0.28.0")
require_version("bitsandbytes>=0.43.0", "To fix: pip install bitsandbytes>=0.43.0")
else:
init_kwargs["device_map"] = {"": get_current_device()}
logger.info("Quantizing model to {} bit.".format(model_args.quantization_bit)) logger.info("Quantizing model to {} bit.".format(model_args.quantization_bit))