uyenlk/SimNPO_forget10_5e-5_Llama-3.2-3B-Instruct_gamma0.25_delta1.0_beta3.5
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: uyenlk/SimNPO_forget10_5e-5_Llama-3.2-3B-Instruct_gamma0.25_delta1.0_beta3.5
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-05-12 18:58:24 +08:00
commit 9935303364
19 changed files with 41773 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
.hydra/overrides.yaml Normal file
View File

@@ -0,0 +1,12 @@
- experiment=unlearn/tofu/default
- model=Llama-3.2-3B-Instruct
- model.model_args.pretrained_model_name_or_path=open-unlearning/tofu_Llama-3.2-3B-Instruct_full
- forget_split=forget10
- retain_split=retain90
- trainer=SimNPO
- trainer.method_args.gamma=0.25
- trainer.method_args.delta=1.0
- trainer.method_args.beta=3.5
- trainer.args.learning_rate=5e-5
- retain_logs_path=saves/eval/tofu_Llama-3.2-3B-Instruct_retain90/TOFU_EVAL.json
- task_name=SimNPO_forget10_5e-5_Llama-3.2-3B-Instruct_gamma0.25_delta1.0_beta3.5