tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

fix: use existing model config

Browse Source
This commit is contained in:
NanoCode012
2025-02-06 00:12:14 +07:00
parent c15ea6b956
commit caa49a9d7d
1 changed files with 1 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/axolotl/cli/convert_linear_attention.py
View File

@@ -49,12 +49,9 @@ def do_linearize(cfg: DictDefault, cli_args: TrainerCliArgs) -> None:
for p in model.parameters(): for p in model.parameters():
p.requires_grad = False p.requires_grad = False
# load config
base_config = load_model_config(cfg)
# convert to linear llama # convert to linear llama
linear_llama_config = LinearLlamaConfig.from_llama( linear_llama_config = LinearLlamaConfig.from_llama(
base_config, cfg.attention_config model.config, cfg.attention_config
) )
model = LinearLlamaForCausalLM.from_llama( model = LinearLlamaForCausalLM.from_llama(
model, config=linear_llama_config, train_attention=True model, config=linear_llama_config, train_attention=True