fsid_non_clustered_0.05_threshold_llama-8b-unlearnt
SimNPO-unlearned checkpoint of meta-llama/Llama-3.1-8B.
Method: SimNPO (Simple Noise-contrastive Policy Optimization)
Forget set threshold: 0.05 memorization fraction
Training: FSDP (2x A100-80GB), gradient checkpointing, 1 epoch
See unlearning-cleanslate/fsid_llama-8b for evaluation artifacts, scores, and validation results.
- Downloads last month
- 3
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for unlearning-cleanslate/fsid_non_clustered_0.05_threshold_llama-8b-unlearnt
Base model
meta-llama/Llama-3.1-8B