RL-Forgetting-Experiments-3/llama-3.2-3b-instruct-code-sft-replay-uniform-lam1-step65

llama3bsftrpuniformlam1

Inference-ready final coding-SFT model for llama3b_sft_rp_uniform_lam1 at optimizer step 65. Training uses llama3b_code_sft_data, order ordered, replay strategy uniform, and replay lambda 1.0.

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