CoolFace
Modelpublic

RL-Forgetting-Experiments-3/qwen2.5-3b-kk-sft-replay-hard-cooldown-acc-lam1-step318

sourceHugging Faceapache-2.0updated 3d agoView on Hugging Face
0likes10downloads
Model Card

Qwen2.5-3B KK replay-SFT: hardcooldownacc, step 318

Status: Intermediate checkpoint. Step 635 was an interrupted invalid save and is intentionally excluded; step 318 is the latest valid model.

Source result: 7209968181.08602-9e7bfe82-135c Source run: qwen3b_kk_ordered_lr1e-5_bs256_seed42_rphard_cooldown_acc_lam1.0_gcr_20260921