starlight-apk/Qwen
1
0
Fork 0
mirror of https://github.com/QwenLM/Qwen.git synced 2026-05-21 00:45:48 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update finetune.py

Browse Source 
Just lora finetuing need low cpu memory usage
This commit is contained in:
梦典
2023-10-18 15:24:45 +08:00
committed by GitHub
parent 67b3f949b6
commit f3d7c69be2
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
finetune.py
View File

@@ -302,7 +302,7 @@ def train():
config=config, config=config,
cache_dir=training_args.cache_dir, cache_dir=training_args.cache_dir,
device_map=device_map, device_map=device_map,
low_cpu_mem_usage=True, low_cpu_mem_usage=True if training_args.use_lora and not lora_args.q_lora,
trust_remote_code=True, trust_remote_code=True,
quantization_config=GPTQConfig( quantization_config=GPTQConfig(
bits=4, disable_exllama=True bits=4, disable_exllama=True