Publish llama3b_code_shuffled_lr1e-5_bs256_seed42/support

runs/llama3b_code_shuffled_lr1e-5_bs256_seed42/init_params.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8874f71944ad57d3c3abc46668aa42c0a05f820fbabfff84d31209416ee7c373
+size 12851071286

runs/llama3b_code_shuffled_lr1e-5_bs256_seed42/sft_config.json

@@ -0,0 +1,44 @@
+{
+  "generation_logs_dir": null,
+  "dataset": "llama3b_code_sft_data",
+  "positives_file": "/mnt/mystorageoutput/projects/rl-forgetting/code-sft-20261027/inputs/llama3b_code_sft_data.jsonl",
+  "model_path": "/mnt/mystorageoutput/projects/rl-forgetting/base_models/Llama-3.2-3B-Instruct",
+  "output_dir": "/mnt/mystorageoutput/projects/rl-forgetting/code-sft-20261027/runs/llama3b_code_shuffled_lr1e-5_bs256_seed42",
+  "lr": 1e-05,
+  "batch_size": 1,
+  "effective_batch_size": 256,
+  "weight_decay": 0.01,
+  "warmup_ratio": 0.0,
+  "schedule": "constant",
+  "max_seq_length": 4096,
+  "seed": 42,
+  "bf16": true,
+  "attn_impl": "auto",
+  "precision": "mixed_bf16",
+  "grad_clip": 0.0,
+  "loss_agg_mode": "token-mean",
+  "ordered": false,
+  "within_step_shuffle": false,
+  "system_prompt_mode": "from_record",
+  "score_threshold": 0.0,
+  "drop_all_pass_groups": false,
+  "max_rl_step": null,
+  "num_epochs": 1,
+  "replay_strategy": "none",
+  "replay_lambda": 0.0,
+  "replay_batch_size": 1,
+  "replay_cooldown_step": 128.0,
+  "replay_ce_tau": 1.0,
+  "replay_ce_init": "insert",
+  "replay_group_size": 8,
+  "replay_seed": null,
+  "num_checkpoints": 10,
+  "save_optimizer_state": true,
+  "prune_old_optimizer": true,
+  "save_optimizer_every_n_ckpts": 1,
+  "snapshot_dtype": "float32",
+  "snapshot_device": "cpu",
+  "metrics_log_every": 1,
+  "resume_from": "latest",
+  "grad_accum_steps": 64
+}