423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-17 04:10:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

bf16 by default, gemma2 attns

Browse Source 
Gemma2 finetuning cannot work until merging https://github.com/huggingface/transformers/pull/31674
This commit is contained in:
hiyouga
2024-06-28 06:00:26 +08:00
parent 64f4337dac
commit 4d35e218b1
3 changed files with 9 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llamafactory/model/patcher.py
View File

@@ -67,7 +67,7 @@ def patch_config(
use_jit_compile = os.environ.get("JIT_COMPILE", "0").lower() in ["true", "1"]
torch.npu.set_compile_mode(jit_compile=use_jit_compile)
configure_attn_implementation(config, model_args)
configure_attn_implementation(config, model_args, is_trainable)
configure_rope(config, model_args, is_trainable)
configure_longlora(config, model_args, is_trainable)
configure_quantization(config, tokenizer, model_args, init_kwargs)