File size: 3,104 Bytes
e5e74c7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
{
  "archive_sha256": "dc7605cecd6c807135e3d83e04496fa96079fef022bed1952c8cf21df5f573da",
  "data_manifest_sha256": "70b144f08930fa2a6b4fde19bb26b8dee4c90bd6fe464cfb23282e7eb986801e",
  "git_revision": "ea63c41cc541fdef8c36da8a85f7b625a2bcec40",
  "plan": {
    "after_completion": "pause_for_evaluation",
    "base": {
      "step": 76294,
      "tokenizer_sha256": "ebae4b4310068ee102f7bf68ac68b79ee1425baa32de3ca8ff8138692529d84e",
      "tokens_seen": 10000007168
    },
    "format": "warpstate-posttrain-plan-v1",
    "handoff": {
      "archive_dir": "rtx6000_40b/work/archives/step-00076294",
      "base_program": "warpstate-train",
      "base_work_dir": "rtx6000_40b/work",
      "data_dir": "rtx6000_40b/work/posttrain/data",
      "poll_seconds": 15,
      "processes": 2,
      "supervisor_socket": "rtx6000_40b/work/control/supervisor.sock",
      "tokenizer": "rtx6000_40b/tokenizer.model",
      "work_dir": "rtx6000_40b/work/posttrain/step-00076294"
    },
    "stage": "sft",
    "training": {
      "bf16_residuals": true,
      "compile_core": true,
      "epochs": 1,
      "eval_every": 100,
      "grad_accum": 4,
      "gradient_checkpointing": false,
      "loss_mode": "compiled-full",
      "lr": 5e-06,
      "max_steps": null,
      "micro_batch": 4,
      "min_lr": 0.0,
      "save_every": 100,
      "seed": 42,
      "sequence_length": 2048,
      "warmup_ratio": 0.03,
      "weight_decay": 0.0,
      "workers": 8
    }
  },
  "source_sha256": {
    "rtx6000_40b/chat.py": "148fd4f20d445ba01723d4c2625372132270884d5c44104f31a3cc13f7c02774",
    "rtx6000_40b/configs/posttrain_10b.json": "a0f8d06e38f72245e046a6a888053c92aa23627256599348f5889886d61df179",
    "rtx6000_40b/ops/checkpoint_writer.py": "3c7fe494e8782802b93a439c05ddd6f416711ca1b3f0f8a196a97a48e37f0216",
    "rtx6000_40b/ops/launch_training.py": "fe9a204023300632766aabe7ff4b0cd22dba5ca7a306040da7ce3b21c9305e08",
    "rtx6000_40b/ops/loss_experiment.py": "12bf7822979692d5106e30544d0ed5d977fb06f556c068ef5f92ddf56ab8376e",
    "rtx6000_40b/ops/posttrain_data.py": "486d2e96e7e99ffd05caf0de7d92068ce47047d4641ede3515681cbb79a9ce10",
    "rtx6000_40b/ops/posttrain_handoff.py": "8f7bb104183bb59802774c5500965d3ac2d1ffd1cb3de18256d075d5414f7c08",
    "rtx6000_40b/ops/posttrain_preflight.py": "4489ae1f37c13e3226d6687147423a9f6a95a0509c135357291ea0a0a9d9664a",
    "rtx6000_40b/ops/residual_experiment.py": "e82aa68e36c76a228156016eb04a02ce1a632a63ac6eda567ec9c4d258aae8f1",
    "rtx6000_40b/ops/training-env.sh": "f5c9d6557d1dd9721e2e4c7bdeef001d994481c863c41bdd4080fd88484aa6f1",
    "rtx6000_40b/posttrain.py": "bd10e97e61b3289e83b7e79c76042bf7f04ce58f4af9067db97ac1fe60d4a411",
    "rtx6000_40b/run_40b.py": "9ad6533d781e376b3b40c16e6c29ef15063d39f917a93800489f17115be15ba0",
    "rtx6000_40b/warpstate/__init__.py": "f386c0880cb499d84159d8205667d250a72d0e8a102f0d7be6af0ba0ea33b22b",
    "rtx6000_40b/warpstate/config.py": "1d716b6c582bab21e8d5fcbaa6d9f3be4f792baf9b6abe826c72422f8eced6ad",
    "rtx6000_40b/warpstate/model.py": "025330ee58cca899695c7631e1a06d2efd006402c61e51106afe6fa2e1d35161"
  }
}