{
 "created_unix": 1790401647,
 "argv": [
  "/workspace/spec_dec_naacl/snapshots/main_code2/specdec/train/train_fused.py",
  "--config",
  "/workspace/spec_dec_naacl/snapshots/main_code2/configs/long/m_long_split_A32.json",
  "--out",
  "/workspace/spec_dec_naacl/outputs/train/long/m_long_split_A32"
 ],
 "config": {
  "name": "m_long_split_A32",
  "description": "M_long: split, L0=48, 32 anchors, 2.0 x 757,624 training rows = 2.02 epochs of the 749,761 usable rows (47352 steps), T1 recipe otherwise",
  "target": {
   "kind": "hf",
   "snapshot": "/workspace/.cache/huggingface/hub/models--Qwen--Qwen3.6-27B/snapshots/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9",
   "dtype": "bfloat16",
   "backend": "triton",
   "fp32_A_log": true
  },
  "device": "cuda",
  "train_data": "outputs/train_data/train",
  "heldout_data": "outputs/train_data/heldout",
  "mask_mode": "split",
  "mask_from_layer": 48,
  "mask_attn_window": null,
  "local_window": 0,
  "rank": 64,
  "alpha": 64,
  "targets": [
   "q_proj",
   "k_proj",
   "v_proj",
   "o_proj",
   "in_proj_qkv",
   "in_proj_z",
   "out_proj",
   "gate_proj",
   "up_proj",
   "down_proj"
  ],
  "block": 15,
  "max_block": 16,
  "max_len": 3072,
  "anchors_per_seq": 32,
  "anchor_sampling": "uniform",
  "hard_candidates": 4,
  "seqs_per_step": 32,
  "steps": 47352,
  "arm_stop_after": null,
  "lr": 0.0003,
  "min_lr_ratio": 0.1,
  "warmup_steps": 200,
  "weight_decay": 0.0,
  "betas": [
   0.9,
   0.999
  ],
  "adam_eps": 1e-08,
  "grad_clip": 1.0,
  "seed": 20260923,
  "eval_every": 1973,
  "eval_rows": 128,
  "eval_anchors_per_seq": 32,
  "eval_seed": 7,
  "eval_walk_rows": 8,
  "eval_walk_chunk": 128,
  "ckpt_every": 1973,
  "keep_last": 24,
  "print_every": 10,
  "freeze_qkv_kv_rows": false,
  "context_scan": "segmented"
 },
 "arm_file": null,
 "stop_after": null,
 "seed": 20260923,
 "git": {
  "head": "2339ed6c03cded5a0a80f4ac072630537c495d50",
  "status": [
   "?? results/long/",
   "?? results/main/ev2/",
   "?? results/main/ev2_shards/",
   "?? results/main/winner.json",
   "?? results/next/"
  ]
 },
 "torch": "2.13.0+cu130",
 "world_size": 4,
 "device": "cuda:0",
 "data_manifest_sha256": "4b69faf9362687f5dfe1f1d46a0dd0d883998800ac05471672a8cf31f8f4b1fa",
 "trainable": {
  "tensors": 201,
  "parameters": 108871680
 },
 "eval_set": {
  "rows": 128,
  "blocks": 4048,
  "sha1": "7b3ba9863a1b785f005cafa614841fbbeef2a9f3"
 }
}