training_config.json

{
 "args": {
  "base": "google/gemma-4-E4B-it",
  "n_per_source": 1000,
  "epochs": 1,
  "lr": 5e-05,
  "head_lr": 0.0,
  "weight_decay": 0.01,
  "lora": 16,
  "accum": 4,
  "holdout": "",
  "suite": "evals/v7/decision-v7",
  "train_sources": "",
  "device": "mps",
  "batch": 2,
  "dtype": "fp32",
  "weights_dtype": "bf16",
  "checkpointing": 1,
  "head_dim": 256,
  "lora_targets": "all",
  "base_revision": "ee0ef6023621cff504d758262d4e04895a5af4a2",
  "p_none": 0.1,
  "p_none_distract": 0.12,
  "p_distract": 0.15,
  "p_none_pair": 0.0,
  "none_pair_max_state": null,
  "synthetic_repeat": 1,
  "public_frac": 1.0,
  "out": "stage2",
  "data": "train-e2b-v06.jsonl",
  "max_state": 5120,
  "extra_suites": "",
  "replay": 0,
  "init_from": "stage1",
  "allow_missing_sources": "",
  "reason": "",
  "row_budget": 0,
  "shared_prefix": 0,
  "length_sort": 0,
  "pass_tokens_max": 0,
  "max_steps": 0,
  "save_every_steps": 0,
  "save_every_minutes": 30.0,
  "resume": 1,
  "stop_after": 0,
  "seed": 0
 },
 "suite_sha256": "a8f50e481b7d90b97da049e0ff6a01cee2f1ed204aed61a8265af0edbb5514d2",
 "base_revision": "ee0ef6023621cff504d758262d4e04895a5af4a2",
 "init_source": {
  "init_from": "stage1 (private: JohnP1/d1a-e4b-runs e4b-it-v06/stage1)",
  "resolved": "stage1 (private: JohnP1/d1a-e4b-runs e4b-it-v06/stage1)",
  "weights_sha256": "bd217060dca2d0ef7d07a103e2bb0c8a4a505502b315b91194a3caa38ba533cd",
  "head_sha256": "e9f13355196677f3bea172ffe26d17bbe4276098f08afaf28db317341b39bd14",
  "tensors": 516
 },
 "holdout": [],
 "sources": {
  "covered": [
   "evals/hard-v1",
   "evals/devtools-v1",
   "evals/documents-v1",
   "evals/d1a/ja-jglue:train",
   "evals/d1a/pr-labels:train",
   "evals/d1a/pr-labels:train-ja",
   "evals/d1a/pr-labels:train-blast",
   "evals/d1a/routing:factory-train",
   "evals/d1a/routing:generic-train",
   "evals/v7/decision-v7"
  ],
  "allowed_missing": [],
  "reason": null
 }
}