RL-Forgetting-Experiments-3/qwen2.5-3b-code-sft-shuffled-lr1e5-step102

q25sftshuffled

Inference-ready final coding-SFT model for q25_sft_shuffled at optimizer step 102. Training uses qwen3b_code_sft_data_s300, order shuffled, replay strategy none, and replay lambda 0.0.

See delivery_manifest.json for immutable source lineage and file checksums.