update int64 | epoch int64 | trajectories int64 | loss float64 | response_tokens int64 | gradient_norm float64 | learning_rate float64 |
|---|---|---|---|---|---|---|
1 | 1 | 64 | 0.419204 | 462,518 | 0.148728 | 0.01 |
2 | 1 | 64 | 0.642034 | 413,410 | 4.909694 | 0.01 |
3 | 1 | 64 | 0.689136 | 435,597 | 6.69772 | 0.01 |
4 | 1 | 64 | 0.457499 | 517,162 | 0.35654 | 0.01 |
5 | 1 | 64 | 0.456273 | 524,676 | 0.365319 | 0.01 |
6 | 1 | 64 | 0.465356 | 452,881 | 0.687765 | 0.01 |
7 | 1 | 64 | 0.458529 | 442,766 | 0.253966 | 0.01 |
8 | 1 | 64 | 0.428442 | 596,289 | 0.100489 | 0.01 |
9 | 1 | 64 | 0.405304 | 434,484 | 0.062905 | 0.01 |
10 | 1 | 64 | 0.415379 | 435,826 | 0.160177 | 0.01 |
11 | 1 | 64 | 0.398037 | 542,257 | 0.045924 | 0.01 |
12 | 1 | 64 | 0.365186 | 462,680 | 0.044405 | 0.01 |
13 | 1 | 64 | 0.448446 | 510,416 | 0.041471 | 0.01 |
14 | 1 | 64 | 0.382729 | 435,477 | 0.05696 | 0.01 |
15 | 1 | 64 | 0.374321 | 456,135 | 0.025664 | 0.01 |
16 | 1 | 64 | 0.356935 | 453,063 | 0.019219 | 0.01 |
17 | 1 | 64 | 0.404136 | 474,911 | 0.026934 | 0.01 |
18 | 1 | 64 | 0.391254 | 437,074 | 0.043352 | 0.01 |
19 | 1 | 64 | 0.457276 | 524,296 | 0.041143 | 0.01 |
20 | 1 | 64 | 0.375842 | 458,031 | 0.025675 | 0.01 |
21 | 1 | 64 | 0.423381 | 461,505 | 0.081694 | 0.01 |
22 | 1 | 64 | 0.38562 | 427,546 | 0.021703 | 0.01 |
23 | 1 | 64 | 0.416312 | 561,888 | 0.01699 | 0.01 |
24 | 1 | 64 | 0.415632 | 475,697 | 0.01397 | 0.01 |
25 | 1 | 64 | 0.388265 | 474,543 | 0.010346 | 0.01 |
26 | 1 | 64 | 0.405619 | 498,297 | 0.013762 | 0.01 |
27 | 1 | 64 | 0.37353 | 435,158 | 0.011463 | 0.01 |
28 | 1 | 64 | 0.413055 | 511,075 | 0.013011 | 0.01 |
29 | 1 | 64 | 0.368889 | 453,190 | 0.010735 | 0.01 |
30 | 1 | 64 | 0.418872 | 462,816 | 0.015375 | 0.01 |
31 | 1 | 64 | 0.397628 | 447,122 | 0.016911 | 0.01 |
32 | 1 | 64 | 0.365315 | 441,619 | 0.00968 | 0.01 |
33 | 1 | 64 | 0.388613 | 472,953 | 0.011921 | 0.01 |
34 | 1 | 64 | 0.383833 | 412,322 | 0.013054 | 0.01 |
35 | 1 | 64 | 0.371644 | 497,952 | 0.022628 | 0.01 |
36 | 1 | 64 | 0.423537 | 528,017 | 0.012487 | 0.01 |
37 | 1 | 64 | 0.399505 | 469,748 | 0.010539 | 0.01 |
38 | 1 | 64 | 0.407621 | 475,393 | 0.008517 | 0.01 |
39 | 1 | 64 | 0.397527 | 444,823 | 0.009621 | 0.01 |
40 | 1 | 64 | 0.399295 | 496,650 | 0.008061 | 0.01 |
41 | 1 | 64 | 0.405584 | 467,326 | 0.015474 | 0.01 |
42 | 1 | 64 | 0.352687 | 398,657 | 0.011008 | 0.01 |
43 | 1 | 64 | 0.372684 | 430,332 | 0.010161 | 0.01 |
44 | 1 | 64 | 0.387567 | 532,123 | 0.011993 | 0.01 |
45 | 1 | 64 | 0.396546 | 411,776 | 0.008705 | 0.01 |
46 | 1 | 64 | 0.380363 | 449,939 | 0.012276 | 0.01 |
47 | 1 | 64 | 0.423177 | 570,533 | 0.018137 | 0.01 |
48 | 1 | 64 | 0.388431 | 457,303 | 0.011538 | 0.01 |
49 | 1 | 64 | 0.417362 | 448,383 | 0.011201 | 0.01 |
50 | 1 | 64 | 0.38554 | 516,431 | 0.00948 | 0.01 |
51 | 1 | 64 | 0.415789 | 520,505 | 0.023001 | 0.01 |
52 | 1 | 64 | 0.363459 | 451,953 | 0.01734 | 0.01 |
53 | 1 | 64 | 0.395707 | 518,020 | 0.016552 | 0.01 |
54 | 1 | 64 | 0.410088 | 500,751 | 0.010672 | 0.01 |
55 | 1 | 64 | 0.406845 | 466,973 | 0.0161 | 0.01 |
56 | 1 | 64 | 0.404936 | 447,500 | 0.029524 | 0.01 |
57 | 1 | 64 | 0.371644 | 424,724 | 0.021357 | 0.01 |
58 | 1 | 64 | 0.426983 | 548,545 | 0.017735 | 0.01 |
59 | 1 | 64 | 0.374751 | 438,724 | 0.011331 | 0.01 |
60 | 1 | 64 | 0.359401 | 477,952 | 0.019521 | 0.01 |
61 | 1 | 64 | 0.399486 | 460,120 | 0.01631 | 0.01 |
62 | 1 | 64 | 0.386035 | 452,197 | 0.014487 | 0.01 |
63 | 1 | 64 | 0.437752 | 613,781 | 0.008207 | 0.01 |
64 | 1 | 64 | 0.401152 | 502,127 | 0.010609 | 0.01 |
65 | 1 | 64 | 0.386558 | 471,854 | 0.010409 | 0.01 |
66 | 1 | 64 | 0.405324 | 489,885 | 0.013628 | 0.01 |
67 | 1 | 64 | 0.375095 | 492,298 | 0.010345 | 0.01 |
68 | 1 | 64 | 0.394647 | 519,511 | 0.010179 | 0.01 |
69 | 1 | 64 | 0.406981 | 490,175 | 0.012707 | 0.01 |
70 | 1 | 64 | 0.38603 | 531,131 | 0.023058 | 0.01 |
71 | 1 | 64 | 0.389573 | 507,033 | 0.014054 | 0.01 |
72 | 1 | 64 | 0.375955 | 425,445 | 0.020209 | 0.01 |
73 | 1 | 64 | 0.410895 | 492,268 | 0.013834 | 0.01 |
74 | 1 | 64 | 0.415171 | 515,315 | 0.012372 | 0.01 |
75 | 1 | 64 | 0.39557 | 519,938 | 0.016653 | 0.01 |
76 | 1 | 64 | 0.466718 | 600,863 | 0.014811 | 0.01 |
77 | 1 | 64 | 0.423029 | 520,390 | 0.010257 | 0.01 |
78 | 1 | 64 | 0.388064 | 534,311 | 0.013522 | 0.01 |
79 | 1 | 64 | 0.380102 | 477,708 | 0.011144 | 0.01 |
80 | 1 | 64 | 0.37128 | 474,417 | 0.011594 | 0.01 |
81 | 1 | 64 | 0.367241 | 416,526 | 0.010787 | 0.01 |
82 | 1 | 64 | 0.399346 | 511,505 | 0.011419 | 0.01 |
83 | 1 | 64 | 0.410075 | 543,718 | 0.009934 | 0.01 |
84 | 1 | 64 | 0.385358 | 496,704 | 0.012045 | 0.01 |
85 | 1 | 64 | 0.375854 | 431,968 | 0.009145 | 0.01 |
86 | 1 | 64 | 0.391811 | 493,059 | 0.009898 | 0.01 |
87 | 1 | 64 | 0.397701 | 515,862 | 0.013377 | 0.01 |
88 | 1 | 64 | 0.379049 | 488,839 | 0.011949 | 0.01 |
89 | 1 | 64 | 0.445705 | 562,492 | 0.015889 | 0.01 |
90 | 1 | 64 | 0.35381 | 471,371 | 0.017311 | 0.01 |
91 | 1 | 64 | 0.409897 | 485,124 | 0.018116 | 0.01 |
92 | 1 | 64 | 0.383856 | 466,528 | 0.011758 | 0.01 |
93 | 1 | 64 | 0.416335 | 521,032 | 0.013301 | 0.01 |
94 | 1 | 64 | 0.374496 | 433,753 | 0.012192 | 0.01 |
95 | 1 | 64 | 0.388367 | 458,877 | 0.012702 | 0.01 |
96 | 1 | 64 | 0.401568 | 472,097 | 0.013012 | 0.01 |
97 | 1 | 64 | 0.36511 | 505,763 | 0.010842 | 0.01 |
98 | 1 | 64 | 0.405309 | 566,901 | 0.010468 | 0.01 |
99 | 1 | 64 | 0.415458 | 488,557 | 0.010002 | 0.01 |
100 | 1 | 64 | 0.388834 | 435,819 | 0.01225 | 0.01 |
Curated agent training backup
Public incremental backup owned by namezz. This preserves the actual training inputs, custom source/configuration snapshot, and selected small adaptation and resume checkpoints. Public backbone weights, package environments, Docker images, and full evaluation container archives are excluded.
Contents
| Folder | Contents |
|---|---|
| data/deepswe | 4,954 frozen teacher episodes, original gzip JSON bytes in five TAR shards |
| data/lego | 1,722 frozen training episodes in four original tar.gz bundles; manifest and schedule |
| data/openthinker | Final corrected JSON training snapshot: 290 episodes, 331 segments |
| code | Custom working-tree source/configuration snapshot and per-file provenance |
| checkpoints | Selected DeepSWE, Lego, OpenThinker and math adaptation/resume artifacts; INDEX.json maps exact source receipts |
| references | Existing HF releases and reused training-source archives; no redundant re-upload |
| provenance | Additional frozen training contracts, batch plans and latest resume receipts |
Restore
Use MANIFEST.json to verify each downloaded file. DeepSWE shards contain individual
.json.gz episode files; SHARD_INDEX.json maps episodes to shards. Lego's
BUNDLE_INDEX.json maps the relocated archives to the original immutable GCS
objects used by its frozen manifest. Expand code/source-and-config.tar.gz into a
fresh directory; SOURCE_MANIFEST.json records every included source file. Retained
model/tokenizer revisions and dependency/configuration records identify public
downloads needed for reconstruction; absolute original cluster paths may require
relocation. Any credential literals removed from source copies are listed by file
and count in code/SNAPSHOT_NOTES.json, without their values.
This is a curated training backup, not a backup of all historical rollouts or every optimizer state. Excluded obsolete template versions must not be mixed with the final OpenThinker training snapshot. Checkpoint inclusion is not a claim of model quality, completed evaluation, or equivalence of configurations across runs. Original dataset and upstream model terms continue to apply.
- Downloads last month
- 239