jbduran commited on
Commit
c94eb54
·
verified ·
1 Parent(s): 4c625d0

remove copy test artifact

Browse files
Files changed (1) hide show
  1. __copytest.json +0 -119
__copytest.json DELETED
@@ -1,119 +0,0 @@
1
- {
2
- "schema_version": 1,
3
- "stage": "base",
4
- "experiment_id": "Think.Unbounded-d32-v2mix-cont",
5
- "branch": {
6
- "parent_experiment_id": "Think.Unbounded-d32",
7
- "parent_step": 5500,
8
- "lr_schedule": "continue",
9
- "load_optimizer": true
10
- },
11
- "datasets": {
12
- "original": {
13
- "adapter": "parquet_shards",
14
- "repo": "jbduran/think-dataset-clean-1930s",
15
- "revision": "45225d95bc15f942be3b4b344738cea1f66e3de8",
16
- "validation_shard": 472,
17
- "num_train_shards": 330,
18
- "download_workers": 4
19
- },
20
- "midtrain_r21": {
21
- "adapter": "parquet_shards",
22
- "repo": "zachnorton03/think-midtrain",
23
- "revision": "5250c497c8c72558a573dbcd216e26ea3c76e6e2",
24
- "subfolder": "mixed/v2/ratio_21/data",
25
- "validation_shard": 65,
26
- "num_train_shards": 65,
27
- "download_workers": 4
28
- },
29
- "midtrain_r45": {
30
- "adapter": "parquet_shards",
31
- "repo": "zachnorton03/think-midtrain",
32
- "revision": "5250c497c8c72558a573dbcd216e26ea3c76e6e2",
33
- "subfolder": "mixed/v2/ratio_45/data",
34
- "validation_shard": 33,
35
- "num_train_shards": 33,
36
- "download_workers": 4
37
- }
38
- },
39
- "mixture_schedule": {
40
- "total_tokens": 20132659200,
41
- "seed_data": 1234,
42
- "max_epochs": 4,
43
- "stages": [
44
- {
45
- "name": "base",
46
- "start_tokens": 0,
47
- "source": "original"
48
- },
49
- {
50
- "name": "injection",
51
- "start_tokens": 14092861440,
52
- "source": "midtrain_r21"
53
- },
54
- {
55
- "name": "decay_mix",
56
- "start_tokens": 18119393280,
57
- "source": "midtrain_r45"
58
- }
59
- ]
60
- },
61
- "tokenizer": {
62
- "mode": "reuse",
63
- "source_experiment_id": "clean1930s-d24-r12-ctx4096-fulltok-v1",
64
- "vocab_size": 32768
65
- },
66
- "pretokenize": {
67
- "enabled": true,
68
- "slack": 1.02,
69
- "require_no_wrap": true,
70
- "val_tokens": 20971520,
71
- "shard_tokens": 100000000,
72
- "tokenizer_threads": 8
73
- },
74
- "training": {
75
- "depth": 32,
76
- "scaling_params": 1677724672,
77
- "target_param_data_ratio": 12,
78
- "max_seq_len": 4096,
79
- "window_pattern": "SSSL",
80
- "muon_momentum": 0.9,
81
- "device_batch_size": 2,
82
- "total_batch_size": 2097152,
83
- "fp8": true,
84
- "fp8_recipe": "tensorwise",
85
- "seed": 42,
86
- "save_every": 500,
87
- "eval_every": 250,
88
- "eval_tokens": 2097152,
89
- "core_metric_every": -1,
90
- "sample_every": -1
91
- },
92
- "artifacts": {
93
- "repo": "jbduran/think.nano",
94
- "keep_local_checkpoints": 1
95
- },
96
- "wandb": {
97
- "entity": "jbduran-thinkingmachinesncsu",
98
- "project": "think.nano",
99
- "name": "Think.Unbounded-d32-v2mix-cont",
100
- "group": "clean1930s-d32",
101
- "tags": [
102
- "think-dataset-clean-1930s",
103
- "d32",
104
- "ratio12",
105
- "ctx4096",
106
- "sssl",
107
- "fp8",
108
- "muon-momentum-constant",
109
- "muon-momentum-0.90",
110
- "midtrain-schedule",
111
- "mix-v2",
112
- "0-21-45",
113
- "midtrain-2.88ep",
114
- "data-repair-continuation"
115
- ]
116
- },
117
- "config_fingerprint": "00af0bde578c1269",
118
- "artifact_path": "experiments/Think.Unbounded-d32-v2mix-cont"
119
- }