Commit ·
8cefc55
1
Parent(s): 9861b2d
checkpoint step2539050 (#340)
Browse files- 29f90f4f34a3f0ce3fda879e60f320b7b5b7c009c38c4249e72adb5a3c9ec044 (bdff6a972ab784a6ae1b864da4be716e2a41ad86)
- e91a3c0764ddbbec40bd6623d122cdcfa805c878f6efd158e3c5275a7da04e9c (f1d6fec539a20158f5cf0b06f327d0ef9d3686e1)
- fec3c255b46ad13e499cb12e5bdd5c044682ab4bc4a28a77783d8f636fee275d (664cfe44b61e4d42781bddc314379fe3cc9ed393)
- cee799cc55948c6e3fe61447ce4b3732fa58741df9f727b5b09056b430545f5b (e61128d6467e77b21418ee970f89d5628a48deda)
- c00b1956eadb8a5ede855bd600c50070f9c363dd03222c3d51ab53bfabd49fa0 (06068d3eb2ac1bc0314f38e8ebf3c34f0afb01f8)
.gitattributes
CHANGED
|
@@ -373,3 +373,4 @@ checkpoints/step2535886_20260830/pretrain_step02535886_from02518657_20260830T065
|
|
| 373 |
checkpoints/step2536677_20260830/pretrain_step02536677_from02518657_20260830T0723Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 374 |
checkpoints/step2537470_20260830/pretrain_step02537470_from02518657_20260830T0753Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 375 |
checkpoints/step2538264_20260830/pretrain_step02538264_from02518657_20260830T0823Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 373 |
checkpoints/step2536677_20260830/pretrain_step02536677_from02518657_20260830T0723Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 374 |
checkpoints/step2537470_20260830/pretrain_step02537470_from02518657_20260830T0753Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 375 |
checkpoints/step2538264_20260830/pretrain_step02538264_from02518657_20260830T0823Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 376 |
+
checkpoints/step2539050_20260830/pretrain_step02539050_from02518657_20260830T0853Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
checkpoints/step2539050_20260830/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c
|
| 3 |
+
size 149
|
checkpoints/step2539050_20260830/serving_candidate.json
ADDED
|
@@ -0,0 +1,546 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"agillm43_provenance": {
|
| 3 |
+
"batch_size": 60,
|
| 4 |
+
"block_size": 2048,
|
| 5 |
+
"checkpoint_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 6 |
+
"checkpoint_type": "full",
|
| 7 |
+
"created_at_iso": "2026-08-30T08:53:30Z",
|
| 8 |
+
"created_at_unix": 1788080010.5575862,
|
| 9 |
+
"created_from_scratch": false,
|
| 10 |
+
"dataset_provenance": {
|
| 11 |
+
"chat": false,
|
| 12 |
+
"dataset_field_text": "text",
|
| 13 |
+
"dataset_hotload_config_path": "/workspace/agillm43_dataset_hotload.json",
|
| 14 |
+
"dataset_hotload_config_sha256": "2762ba666c280207bdb2293b6d382ef8077d167de79609e5bcc35d4499fd1527",
|
| 15 |
+
"dataset_hotload_config_stamp": [
|
| 16 |
+
73,
|
| 17 |
+
2196426857,
|
| 18 |
+
1787483694198261655,
|
| 19 |
+
1281
|
| 20 |
+
],
|
| 21 |
+
"dataset_hotload_config_used": true,
|
| 22 |
+
"dataset_hotload_enabled": true,
|
| 23 |
+
"dataset_hotload_generation": 4,
|
| 24 |
+
"dataset_hotload_schema": "agillm.dataset_hotload.config.v2",
|
| 25 |
+
"dataset_revision_map_sha256": "8361995a504a2365136b20e1a879ba78944a06309f13f3f013bbd0cdfc72225b",
|
| 26 |
+
"dataset_revisions": {
|
| 27 |
+
"EleutherAI/proof-pile-2": "901a9273a770e9d4138c5ddd91802f9c5c6cdc4b",
|
| 28 |
+
"HuggingFaceFW/fineweb": "9bb295ddab0e05d785b879661af7260fed5140fc",
|
| 29 |
+
"HuggingFaceFW/fineweb-edu": "87f09149ef4734204d70ed1d046ddc9ca3f2b8f9",
|
| 30 |
+
"HuggingFaceTB/cosmopedia": "0ae6ec63f91742bd2d1eaef4f02232c55d719385",
|
| 31 |
+
"HuggingFaceTB/finemath": "e92b25a616738fe95dc186b64dfb19f9c8525594",
|
| 32 |
+
"HuggingFaceTB/smollm-corpus": "3ba9d605774198c5868892d7a8deda78031a781f",
|
| 33 |
+
"Skylion007/openwebtext": "79d93d786212f7344586290adb811d4ae6a1762c",
|
| 34 |
+
"allenai/c4": "1588ec454efa1a09f29cd18ddd04fe05fc8653a2",
|
| 35 |
+
"allenai/dolma": "7f48140530a023e9ea4c5cfb141160922727d4d3",
|
| 36 |
+
"allenai/dolmino-mix-1124": "a319f19eef1e257417b11ea8c30da266ae175557",
|
| 37 |
+
"codeparrot/codeparrot-clean": "35a59fb025bc0a102f7d96eac09d145b896d487b",
|
| 38 |
+
"mlfoundations/dclm-baseline-1.0": "a3b142c183aebe5af344955ae20836eb34dcf69b",
|
| 39 |
+
"wikimedia/wikipedia": "b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
|
| 40 |
+
},
|
| 41 |
+
"dolma_payload_integrity": "external_url_manifest_pinned_payload_hashes_unavailable",
|
| 42 |
+
"dolma_url_list_sha256": "bc6ef505be920c94bb9e7147445638f606fdc3f5af622cd1d887dc48e21a2d5b",
|
| 43 |
+
"dolma_url_manifest_raw_sha256": "6fe8c40d51e5bebc9303453d7f533b232d69f7ae122123621a82621b1cd779b7",
|
| 44 |
+
"has_language_mix": true,
|
| 45 |
+
"has_numeracy": false,
|
| 46 |
+
"hot_config_expected_sha256": "",
|
| 47 |
+
"hot_config_has_append": true,
|
| 48 |
+
"hot_config_has_datasets": false,
|
| 49 |
+
"hot_config_mtime": 1787722590.529335,
|
| 50 |
+
"hot_config_path": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
|
| 51 |
+
"hot_config_used": true,
|
| 52 |
+
"hot_reload_capable": true,
|
| 53 |
+
"phase": "pretrain",
|
| 54 |
+
"proof_pile_lfs_manifest_sha256": "d020dfb796b02a75564e4e11d9e03185a55b3f9c6639f87830cfbe10737ed329",
|
| 55 |
+
"proof_pile_paths_sha256": "8a30018017e3b3d9a9bcfa06a3c404ccd113297b0b1d0c48e2e2aadf113e52b0",
|
| 56 |
+
"schema": "agillm.dataset_provenance.v1",
|
| 57 |
+
"sequence_prefetch_depth": 256,
|
| 58 |
+
"source_count": 15,
|
| 59 |
+
"source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25,HuggingFaceTB/finemath:finemath-4plus|0.7,HuggingFaceTB/finemath:infiwebmath-4plus|0.35,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8,allenai/dolmino-mix-1124|1,mlfoundations/dclm-baseline-1.0|1,HuggingFaceFW/fineweb-edu:sample-100BT|1.6",
|
| 60 |
+
"source_requested": "HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25",
|
| 61 |
+
"source_specs": [
|
| 62 |
+
"HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50",
|
| 63 |
+
"wikimedia/wikipedia:20231101.en|1.10",
|
| 64 |
+
"allenai/c4:en|0.40",
|
| 65 |
+
"Skylion007/openwebtext|0.35",
|
| 66 |
+
"HuggingFaceTB/cosmopedia:web_samples_v2|1.30",
|
| 67 |
+
"HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30",
|
| 68 |
+
"EleutherAI/proof-pile-2:all|1.50",
|
| 69 |
+
"allenai/dolma:v1_6-sample|0.50",
|
| 70 |
+
"codeparrot/codeparrot-clean|1.25",
|
| 71 |
+
"HuggingFaceTB/finemath:finemath-4plus|0.7",
|
| 72 |
+
"HuggingFaceTB/finemath:infiwebmath-4plus|0.35",
|
| 73 |
+
"HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8",
|
| 74 |
+
"allenai/dolmino-mix-1124|1",
|
| 75 |
+
"mlfoundations/dclm-baseline-1.0|1",
|
| 76 |
+
"HuggingFaceFW/fineweb-edu:sample-100BT|1.6"
|
| 77 |
+
],
|
| 78 |
+
"val_source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25,HuggingFaceTB/finemath:finemath-4plus|0.7,HuggingFaceTB/finemath:infiwebmath-4plus|0.35,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8,allenai/dolmino-mix-1124|1,mlfoundations/dclm-baseline-1.0|1,HuggingFaceFW/fineweb-edu:sample-100BT|1.6",
|
| 79 |
+
"val_source_requested": ""
|
| 80 |
+
},
|
| 81 |
+
"effective_global_step": 5057707,
|
| 82 |
+
"effective_seen_tok": 245714677760,
|
| 83 |
+
"global_origin_seen_tok": 0,
|
| 84 |
+
"global_origin_step": 2518657,
|
| 85 |
+
"gpu": {
|
| 86 |
+
"allocated_gb": 6.83,
|
| 87 |
+
"peak_allocated_gb": 18.92,
|
| 88 |
+
"reserved_gb": 22.53
|
| 89 |
+
},
|
| 90 |
+
"hostname": "ead6505f92c4",
|
| 91 |
+
"lane": "pretrain",
|
| 92 |
+
"launch": {
|
| 93 |
+
"argv": [
|
| 94 |
+
"/workspace/agillm43_production_v14/runtime/agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py",
|
| 95 |
+
"train",
|
| 96 |
+
"--preset",
|
| 97 |
+
"agillm4_floor",
|
| 98 |
+
"--block",
|
| 99 |
+
"2048",
|
| 100 |
+
"--source",
|
| 101 |
+
"HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25",
|
| 102 |
+
"--nat_mask_token_id",
|
| 103 |
+
"2",
|
| 104 |
+
"--tie_weights",
|
| 105 |
+
"--tie_kv",
|
| 106 |
+
"--moe_ffn",
|
| 107 |
+
"--moe_experts",
|
| 108 |
+
"2",
|
| 109 |
+
"--moe_top_k",
|
| 110 |
+
"1",
|
| 111 |
+
"--moe_mlp_mult",
|
| 112 |
+
"4",
|
| 113 |
+
"--moe_shared_experts",
|
| 114 |
+
"1",
|
| 115 |
+
"--moe_shared_mlp_mult",
|
| 116 |
+
"2",
|
| 117 |
+
"--moe_aux_coef",
|
| 118 |
+
"0.01",
|
| 119 |
+
"--moe_z_coef",
|
| 120 |
+
"0.001",
|
| 121 |
+
"--amp",
|
| 122 |
+
"--grad_checkpoint",
|
| 123 |
+
"--attn_backend",
|
| 124 |
+
"sdpa",
|
| 125 |
+
"--sublinear_window",
|
| 126 |
+
"128",
|
| 127 |
+
"--sublinear_stride",
|
| 128 |
+
"128",
|
| 129 |
+
"--sublinear_max_anchors",
|
| 130 |
+
"128",
|
| 131 |
+
"--sublinear_chunk",
|
| 132 |
+
"128",
|
| 133 |
+
"--sublinear_sinks",
|
| 134 |
+
"4",
|
| 135 |
+
"--sublinear_recent_anchors",
|
| 136 |
+
"64",
|
| 137 |
+
"--no-sublinear_pooled_landmarks",
|
| 138 |
+
"--anchor_stride",
|
| 139 |
+
"256",
|
| 140 |
+
"--anchor_max",
|
| 141 |
+
"2048",
|
| 142 |
+
"--anchor_position",
|
| 143 |
+
"-1",
|
| 144 |
+
"--alibi_mode",
|
| 145 |
+
"corrected",
|
| 146 |
+
"--alibi_scale",
|
| 147 |
+
"0.0",
|
| 148 |
+
"--optimizer",
|
| 149 |
+
"adamw8bit",
|
| 150 |
+
"--weight_decay",
|
| 151 |
+
"0.0",
|
| 152 |
+
"--lr_decay",
|
| 153 |
+
"cosine",
|
| 154 |
+
"--lr_decay_tokens",
|
| 155 |
+
"49000954880",
|
| 156 |
+
"--lr_min_mult",
|
| 157 |
+
"0.2",
|
| 158 |
+
"--lr_schedule_reset_on_resume",
|
| 159 |
+
"--lr_warmup_tokens",
|
| 160 |
+
"10000000",
|
| 161 |
+
"--lr_warmup_min_mult",
|
| 162 |
+
"0.1",
|
| 163 |
+
"--dblock",
|
| 164 |
+
"--dblock_blocks",
|
| 165 |
+
"14",
|
| 166 |
+
"--dblock_schedule",
|
| 167 |
+
"roundrobin",
|
| 168 |
+
"--dblock_router",
|
| 169 |
+
"shadow",
|
| 170 |
+
"--dblock_router_blend",
|
| 171 |
+
"0.0",
|
| 172 |
+
"--dblock_router_ramp_steps",
|
| 173 |
+
"64",
|
| 174 |
+
"--dblock_warmup_steps",
|
| 175 |
+
"14",
|
| 176 |
+
"--dblock_sublayer_mode",
|
| 177 |
+
"off",
|
| 178 |
+
"--dblock_checkpoint_skip_tail",
|
| 179 |
+
"0",
|
| 180 |
+
"--dblock_loop_layers",
|
| 181 |
+
"0",
|
| 182 |
+
"--dblock_loop_start",
|
| 183 |
+
"0",
|
| 184 |
+
"--dblock_loop_cond_scale",
|
| 185 |
+
"1.0",
|
| 186 |
+
"--dblock_sigma_curriculum_steps",
|
| 187 |
+
"2000",
|
| 188 |
+
"--dblock_sigma_sampling",
|
| 189 |
+
"lognormal",
|
| 190 |
+
"--dblock_sigma_stratified",
|
| 191 |
+
"--dblock_sigma_min",
|
| 192 |
+
"0.002",
|
| 193 |
+
"--dblock_sigma_max",
|
| 194 |
+
"2.0",
|
| 195 |
+
"--dblock_sigma_pmean",
|
| 196 |
+
"-1.2",
|
| 197 |
+
"--dblock_sigma_pstd",
|
| 198 |
+
"1.2",
|
| 199 |
+
"--dblock_edm_wmax",
|
| 200 |
+
"5.0",
|
| 201 |
+
"--dblock_ar_prob",
|
| 202 |
+
"0.8",
|
| 203 |
+
"--dblock_sat_prob",
|
| 204 |
+
"0.1",
|
| 205 |
+
"--dblock_nat_prob",
|
| 206 |
+
"0.1",
|
| 207 |
+
"--dblock_ar_loss_tokens",
|
| 208 |
+
"8192",
|
| 209 |
+
"--dblock_sat_loss_tokens",
|
| 210 |
+
"4096",
|
| 211 |
+
"--dblock_nat_loss_tokens",
|
| 212 |
+
"4096",
|
| 213 |
+
"--dblock_nat_embed_noise_mode",
|
| 214 |
+
"off",
|
| 215 |
+
"--dblock_nat_embed_noise_scale",
|
| 216 |
+
"1.0",
|
| 217 |
+
"--nat_loss_weight",
|
| 218 |
+
"1.0",
|
| 219 |
+
"--nat_mask_ratio",
|
| 220 |
+
"0.5",
|
| 221 |
+
"--nat_max_tokens",
|
| 222 |
+
"0",
|
| 223 |
+
"--nat_span_mask_prob",
|
| 224 |
+
"0.35",
|
| 225 |
+
"--nat_suffix_mask_prob",
|
| 226 |
+
"0.20",
|
| 227 |
+
"--nat_span_max_tokens",
|
| 228 |
+
"0",
|
| 229 |
+
"--loss_spike_skip",
|
| 230 |
+
"0.0",
|
| 231 |
+
"--sat_every",
|
| 232 |
+
"1",
|
| 233 |
+
"--nat_every",
|
| 234 |
+
"1",
|
| 235 |
+
"--dblock_fullstack_anchor_spike_skip",
|
| 236 |
+
"0.0",
|
| 237 |
+
"--dblock_fullstack_anchor_max_ce",
|
| 238 |
+
"0.0",
|
| 239 |
+
"--dblock_anchor_softcap_mult",
|
| 240 |
+
"4.0",
|
| 241 |
+
"--dblock_anchor_softcap_min_ce",
|
| 242 |
+
"25.0",
|
| 243 |
+
"--dblock_anchor_softcap_max_ce",
|
| 244 |
+
"50.0",
|
| 245 |
+
"--dblock_fullstack_ar_every",
|
| 246 |
+
"3",
|
| 247 |
+
"--dblock_fullstack_ar_offset",
|
| 248 |
+
"0",
|
| 249 |
+
"--dblock_fullstack_ar_tokens",
|
| 250 |
+
"256",
|
| 251 |
+
"--dblock_fullstack_sat_every",
|
| 252 |
+
"3",
|
| 253 |
+
"--dblock_fullstack_sat_offset",
|
| 254 |
+
"1",
|
| 255 |
+
"--dblock_fullstack_sat_tokens",
|
| 256 |
+
"128",
|
| 257 |
+
"--dblock_fullstack_nat_every",
|
| 258 |
+
"3",
|
| 259 |
+
"--dblock_fullstack_nat_offset",
|
| 260 |
+
"2",
|
| 261 |
+
"--dblock_fullstack_nat_tokens",
|
| 262 |
+
"128",
|
| 263 |
+
"--dblock_fullstack_nat_mask_id",
|
| 264 |
+
"-1",
|
| 265 |
+
"--ckpt_codec",
|
| 266 |
+
"block-sharded-zstd",
|
| 267 |
+
"--async_update_every_steps",
|
| 268 |
+
"0",
|
| 269 |
+
"--dblock_router_hidden",
|
| 270 |
+
"64",
|
| 271 |
+
"--dblock_router_heads",
|
| 272 |
+
"4",
|
| 273 |
+
"--dblock_router_layers",
|
| 274 |
+
"2",
|
| 275 |
+
"--dblock_router_lr",
|
| 276 |
+
"0.0005",
|
| 277 |
+
"--dblock_explore",
|
| 278 |
+
"0.08",
|
| 279 |
+
"--dblock_max_stale_steps",
|
| 280 |
+
"64",
|
| 281 |
+
"--dblock_max_count_skew",
|
| 282 |
+
"1.35",
|
| 283 |
+
"--dblock_stale_bonus",
|
| 284 |
+
"0.35",
|
| 285 |
+
"--dblock_undertrain_bonus",
|
| 286 |
+
"0.25",
|
| 287 |
+
"--dblock_activation_offload_min_mb",
|
| 288 |
+
"1.0",
|
| 289 |
+
"--nat_document_boundary_aware",
|
| 290 |
+
"--dblock_fullstack_anchor_deterministic_eval",
|
| 291 |
+
"--dblock_fullstack_anchor_spike_retry_limit",
|
| 292 |
+
"3",
|
| 293 |
+
"--dblock_local_spike_retry_limit",
|
| 294 |
+
"3",
|
| 295 |
+
"--dblock_fullstack_nat_no_valid_crop_limit",
|
| 296 |
+
"8",
|
| 297 |
+
"--target_tokens",
|
| 298 |
+
"400000000000",
|
| 299 |
+
"--lr_schedule_reanchor_on_resume",
|
| 300 |
+
"--save_every_sec",
|
| 301 |
+
"1800",
|
| 302 |
+
"--delta_every_steps",
|
| 303 |
+
"0",
|
| 304 |
+
"--delta_every_sec",
|
| 305 |
+
"0",
|
| 306 |
+
"--delta_max_keep",
|
| 307 |
+
"0",
|
| 308 |
+
"--max_ckpts",
|
| 309 |
+
"2",
|
| 310 |
+
"--disk_free_floor_gb",
|
| 311 |
+
"12.0",
|
| 312 |
+
"--heartbeat_every_sec",
|
| 313 |
+
"30",
|
| 314 |
+
"--dblock_fullstack_ar_weight",
|
| 315 |
+
"1.0",
|
| 316 |
+
"--dblock_fullstack_sat_weight",
|
| 317 |
+
"0.1",
|
| 318 |
+
"--dblock_fullstack_nat_weight",
|
| 319 |
+
"0.1",
|
| 320 |
+
"--dblock_local_aux_warn_ce",
|
| 321 |
+
"0.0",
|
| 322 |
+
"--dblock_local_aux_max_ce",
|
| 323 |
+
"0.0",
|
| 324 |
+
"--lr_schedule_origin_tokens",
|
| 325 |
+
"85002387456",
|
| 326 |
+
"--dblock_stop_after_commits",
|
| 327 |
+
"0",
|
| 328 |
+
"--no-oom_auto_backoff",
|
| 329 |
+
"--require_exact_resume_state",
|
| 330 |
+
"--dblock_objective_mode",
|
| 331 |
+
"committed_9_1_1",
|
| 332 |
+
"--dblock_ar_weight",
|
| 333 |
+
"0.25",
|
| 334 |
+
"--dblock_sat_weight",
|
| 335 |
+
"0.25",
|
| 336 |
+
"--dblock_nat_weight",
|
| 337 |
+
"0.25",
|
| 338 |
+
"--dblock_aux_grad_ratio",
|
| 339 |
+
"0.5",
|
| 340 |
+
"--dblock_min_ar_share",
|
| 341 |
+
"0.8",
|
| 342 |
+
"--lr_core",
|
| 343 |
+
"2e-7",
|
| 344 |
+
"--lr_head",
|
| 345 |
+
"2e-7",
|
| 346 |
+
"--batch_size",
|
| 347 |
+
"60",
|
| 348 |
+
"--oom_known_safe_batch",
|
| 349 |
+
"56",
|
| 350 |
+
"--dblock_checkpoint_stride",
|
| 351 |
+
"2",
|
| 352 |
+
"--empty_cache_every_steps",
|
| 353 |
+
"0",
|
| 354 |
+
"--dblock_log_every",
|
| 355 |
+
"25",
|
| 356 |
+
"--cuda_max_reserved_mib",
|
| 357 |
+
"0",
|
| 358 |
+
"--cuda_min_free_mib",
|
| 359 |
+
"0",
|
| 360 |
+
"--weight_accel_mode",
|
| 361 |
+
"shadow",
|
| 362 |
+
"--weight_accel_dir",
|
| 363 |
+
"/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/weight_trajectory_accel",
|
| 364 |
+
"--weight_accel_every",
|
| 365 |
+
"1",
|
| 366 |
+
"--weight_accel_train_every",
|
| 367 |
+
"4",
|
| 368 |
+
"--weight_accel_record_every",
|
| 369 |
+
"14",
|
| 370 |
+
"--weight_accel_state_every",
|
| 371 |
+
"256",
|
| 372 |
+
"--resume",
|
| 373 |
+
"/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt",
|
| 374 |
+
"--save_dir",
|
| 375 |
+
"/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B",
|
| 376 |
+
"--data_seed",
|
| 377 |
+
"2518699"
|
| 378 |
+
],
|
| 379 |
+
"argv_string": "/workspace/agillm43_production_v14/runtime/agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py train --preset agillm4_floor --block 2048 --source HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25 --nat_mask_token_id 2 --tie_weights --tie_kv --moe_ffn --moe_experts 2 --moe_top_k 1 --moe_mlp_mult 4 --moe_shared_experts 1 --moe_shared_mlp_mult 2 --moe_aux_coef 0.01 --moe_z_coef 0.001 --amp --grad_checkpoint --attn_backend sdpa --sublinear_window 128 --sublinear_stride 128 --sublinear_max_anchors 128 --sublinear_chunk 128 --sublinear_sinks 4 --sublinear_recent_anchors 64 --no-sublinear_pooled_landmarks --anchor_stride 256 --anchor_max 2048 --anchor_position -1 --alibi_mode corrected --alibi_scale 0.0 --optimizer adamw8bit --weight_decay 0.0 --lr_decay cosine --lr_decay_tokens 49000954880 --lr_min_mult 0.2 --lr_schedule_reset_on_resume --lr_warmup_tokens 10000000 --lr_warmup_min_mult 0.1 --dblock --dblock_blocks 14 --dblock_schedule roundrobin --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --dblock_sublayer_mode off --dblock_checkpoint_skip_tail 0 --dblock_loop_layers 0 --dblock_loop_start 0 --dblock_loop_cond_scale 1.0 --dblock_sigma_curriculum_steps 2000 --dblock_sigma_sampling lognormal --dblock_sigma_stratified --dblock_sigma_min 0.002 --dblock_sigma_max 2.0 --dblock_sigma_pmean -1.2 --dblock_sigma_pstd 1.2 --dblock_edm_wmax 5.0 --dblock_ar_prob 0.8 --dblock_sat_prob 0.1 --dblock_nat_prob 0.1 --dblock_ar_loss_tokens 8192 --dblock_sat_loss_tokens 4096 --dblock_nat_loss_tokens 4096 --dblock_nat_embed_noise_mode off --dblock_nat_embed_noise_scale 1.0 --nat_loss_weight 1.0 --nat_mask_ratio 0.5 --nat_max_tokens 0 --nat_span_mask_prob 0.35 --nat_suffix_mask_prob 0.20 --nat_span_max_tokens 0 --loss_spike_skip 0.0 --sat_every 1 --nat_every 1 --dblock_fullstack_anchor_spike_skip 0.0 --dblock_fullstack_anchor_max_ce 0.0 --dblock_anchor_softcap_mult 4.0 --dblock_anchor_softcap_min_ce 25.0 --dblock_anchor_softcap_max_ce 50.0 --dblock_fullstack_ar_every 3 --dblock_fullstack_ar_offset 0 --dblock_fullstack_ar_tokens 256 --dblock_fullstack_sat_every 3 --dblock_fullstack_sat_offset 1 --dblock_fullstack_sat_tokens 128 --dblock_fullstack_nat_every 3 --dblock_fullstack_nat_offset 2 --dblock_fullstack_nat_tokens 128 --dblock_fullstack_nat_mask_id -1 --ckpt_codec block-sharded-zstd --async_update_every_steps 0 --dblock_router_hidden 64 --dblock_router_heads 4 --dblock_router_layers 2 --dblock_router_lr 0.0005 --dblock_explore 0.08 --dblock_max_stale_steps 64 --dblock_max_count_skew 1.35 --dblock_stale_bonus 0.35 --dblock_undertrain_bonus 0.25 --dblock_activation_offload_min_mb 1.0 --nat_document_boundary_aware --dblock_fullstack_anchor_deterministic_eval --dblock_fullstack_anchor_spike_retry_limit 3 --dblock_local_spike_retry_limit 3 --dblock_fullstack_nat_no_valid_crop_limit 8 --target_tokens 400000000000 --lr_schedule_reanchor_on_resume --save_every_sec 1800 --delta_every_steps 0 --delta_every_sec 0 --delta_max_keep 0 --max_ckpts 2 --disk_free_floor_gb 12.0 --heartbeat_every_sec 30 --dblock_fullstack_ar_weight 1.0 --dblock_fullstack_sat_weight 0.1 --dblock_fullstack_nat_weight 0.1 --dblock_local_aux_warn_ce 0.0 --dblock_local_aux_max_ce 0.0 --lr_schedule_origin_tokens 85002387456 --dblock_stop_after_commits 0 --no-oom_auto_backoff --require_exact_resume_state --dblock_objective_mode committed_9_1_1 --dblock_ar_weight 0.25 --dblock_sat_weight 0.25 --dblock_nat_weight 0.25 --dblock_aux_grad_ratio 0.5 --dblock_min_ar_share 0.8 --lr_core 2e-7 --lr_head 2e-7 --batch_size 60 --oom_known_safe_batch 56 --dblock_checkpoint_stride 2 --empty_cache_every_steps 0 --dblock_log_every 25 --cuda_max_reserved_mib 0 --cuda_min_free_mib 0 --weight_accel_mode shadow --weight_accel_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/weight_trajectory_accel --weight_accel_every 1 --weight_accel_train_every 4 --weight_accel_record_every 14 --weight_accel_state_every 256 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2518699",
|
| 380 |
+
"cwd": "/workspace/agillm43_production_v14/runtime",
|
| 381 |
+
"env": {
|
| 382 |
+
"AGILLM43_DECOMPRESS_CACHE": "0",
|
| 383 |
+
"AGILLM43_SEQUENCE_PREFETCH_DEPTH": "256",
|
| 384 |
+
"AGILLM43_SINGLEFILE_PROFILE": "intelligence-v25-satvar-pretrain",
|
| 385 |
+
"AGILLM_DATASETS_STREAMING_READ_MAX_RETRIES": "2",
|
| 386 |
+
"AGILLM_DATASETS_STREAMING_READ_RETRY_INTERVAL": "1",
|
| 387 |
+
"AGILLM_DATASET_AGENT_ROUTER": "0",
|
| 388 |
+
"AGILLM_DATASET_HOTLOAD_CONFIG": "/workspace/agillm43_dataset_hotload.json",
|
| 389 |
+
"AGILLM_DATASET_HOT_RELOAD_CHECK_TOKENS": "<redacted>",
|
| 390 |
+
"AGILLM_DATASET_HOT_RELOAD_SEC": "5",
|
| 391 |
+
"AGILLM_DATASET_NN_ROUTER": "1",
|
| 392 |
+
"AGILLM_DATASET_ROUTER_BLEND": "0.25",
|
| 393 |
+
"AGILLM_DATASET_ROUTER_EXPLORE": "0.03",
|
| 394 |
+
"AGILLM_DATASET_ROUTER_LR": "0.005",
|
| 395 |
+
"AGILLM_DATASET_ROUTER_MIN_SCORE": "0.15",
|
| 396 |
+
"AGILLM_DATASET_ROUTER_SHARPNESS": "2.0",
|
| 397 |
+
"AGILLM_ENABLE_CUPY": "0",
|
| 398 |
+
"AGILLM_HF_HUB_DOWNLOAD_TIMEOUT": "30",
|
| 399 |
+
"AGILLM_HF_TRANSPORT": "geth_hf_relay",
|
| 400 |
+
"AGILLM_HOT_CONFIG": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
|
| 401 |
+
"AGILLM_HOT_CONFIG_SHA256": "",
|
| 402 |
+
"AGILLM_MAX_EXAMPLE_CHARS": "32768",
|
| 403 |
+
"AGILLM_MAX_EXAMPLE_TOKENS": "<redacted>",
|
| 404 |
+
"AGILLM_PARENT_POINTER_SHA256": "0fb3c1b33756f72da78f5419606307366223602929633a581fb7431b6057dd69",
|
| 405 |
+
"AGILLM_REMOTE_SHARDS_PER_ITER": "4",
|
| 406 |
+
"AGILLM_RUN_ID": "singlefile_intelligence_v25_20260829T195836Z_from_step000002518657",
|
| 407 |
+
"AGILLM_STREAM_NEXT_TIMEOUT_SEC": "60",
|
| 408 |
+
"AGILLM_STREAM_OPEN_TIMEOUT_SEC": "120",
|
| 409 |
+
"AGILLM_STREAM_SOCKET_TIMEOUT_SEC": "30",
|
| 410 |
+
"AGILLM_STREAM_SOURCE_FATAL_COOLDOWN_SEC": "600",
|
| 411 |
+
"AGILLM_STREAM_SOURCE_MAX_COOLDOWN_SEC": "120",
|
| 412 |
+
"AGILLM_SYNTHETIC_TOKENIZER": "<redacted>",
|
| 413 |
+
"CUDA_MODULE_LOADING": "LAZY",
|
| 414 |
+
"CUDA_VERSION": "12.1.1",
|
| 415 |
+
"CUDA_VISIBLE_DEVICES": "0",
|
| 416 |
+
"HF_HUB_DOWNLOAD_TIMEOUT": "30",
|
| 417 |
+
"HF_HUB_ETAG_TIMEOUT": "10",
|
| 418 |
+
"PYTORCH_CUDA_ALLOC_CONF": "backend:cudaMallocAsync",
|
| 419 |
+
"PYTORCH_VERSION": "2.2.1"
|
| 420 |
+
},
|
| 421 |
+
"parent_proc_cmdline": "/opt/conda/bin/python -u /workspace/agillm43_production_v14/runtime/agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py supervise --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --log /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/logs/singlefile_intelligence_v25_satvar_pretrain.log --sleep_sec 15 --profile normal",
|
| 422 |
+
"pid": 3402475,
|
| 423 |
+
"ppid": 3560703,
|
| 424 |
+
"proc_cmdline": "/opt/conda/bin/python -u /workspace/agillm43_production_v14/runtime/agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py train --preset agillm4_floor --block 2048 --source HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25 --nat_mask_token_id 2 --tie_weights --tie_kv --moe_ffn --moe_experts 2 --moe_top_k 1 --moe_mlp_mult 4 --moe_shared_experts 1 --moe_shared_mlp_mult 2 --moe_aux_coef 0.01 --moe_z_coef 0.001 --amp --grad_checkpoint --attn_backend sdpa --sublinear_window 128 --sublinear_stride 128 --sublinear_max_anchors 128 --sublinear_chunk 128 --sublinear_sinks 4 --sublinear_recent_anchors 64 --no-sublinear_pooled_landmarks --anchor_stride 256 --anchor_max 2048 --anchor_position -1 --alibi_mode corrected --alibi_scale 0.0 --optimizer adamw8bit --weight_decay 0.0 --lr_decay cosine --lr_decay_tokens 49000954880 --lr_min_mult 0.2 --lr_schedule_reset_on_resume --lr_warmup_tokens 10000000 --lr_warmup_min_mult 0.1 --dblock --dblock_blocks 14 --dblock_schedule roundrobin --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --dblock_sublayer_mode off --dblock_checkpoint_skip_tail 0 --dblock_loop_layers 0 --dblock_loop_start 0 --dblock_loop_cond_scale 1.0 --dblock_sigma_curriculum_steps 2000 --dblock_sigma_sampling lognormal --dblock_sigma_stratified --dblock_sigma_min 0.002 --dblock_sigma_max 2.0 --dblock_sigma_pmean -1.2 --dblock_sigma_pstd 1.2 --dblock_edm_wmax 5.0 --dblock_ar_prob 0.8 --dblock_sat_prob 0.1 --dblock_nat_prob 0.1 --dblock_ar_loss_tokens 8192 --dblock_sat_loss_tokens 4096 --dblock_nat_loss_tokens 4096 --dblock_nat_embed_noise_mode off --dblock_nat_embed_noise_scale 1.0 --nat_loss_weight 1.0 --nat_mask_ratio 0.5 --nat_max_tokens 0 --nat_span_mask_prob 0.35 --nat_suffix_mask_prob 0.20 --nat_span_max_tokens 0 --loss_spike_skip 0.0 --sat_every 1 --nat_every 1 --dblock_fullstack_anchor_spike_skip 0.0 --dblock_fullstack_anchor_max_ce 0.0 --dblock_anchor_softcap_mult 4.0 --dblock_anchor_softcap_min_ce 25.0 --dblock_anchor_softcap_max_ce 50.0 --dblock_fullstack_ar_every 3 --dblock_fullstack_ar_offset 0 --dblock_fullstack_ar_tokens 256 --dblock_fullstack_sat_every 3 --dblock_fullstack_sat_offset 1 --dblock_fullstack_sat_tokens 128 --dblock_fullstack_nat_every 3 --dblock_fullstack_nat_offset 2 --dblock_fullstack_nat_tokens 128 --dblock_fullstack_nat_mask_id -1 --ckpt_codec block-sharded-zstd --async_update_every_steps 0 --dblock_router_hidden 64 --dblock_router_heads 4 --dblock_router_layers 2 --dblock_router_lr 0.0005 --dblock_explore 0.08 --dblock_max_stale_steps 64 --dblock_max_count_skew 1.35 --dblock_stale_bonus 0.35 --dblock_undertrain_bonus 0.25 --dblock_activation_offload_min_mb 1.0 --nat_document_boundary_aware --dblock_fullstack_anchor_deterministic_eval --dblock_fullstack_anchor_spike_retry_limit 3 --dblock_local_spike_retry_limit 3 --dblock_fullstack_nat_no_valid_crop_limit 8 --target_tokens 400000000000 --lr_schedule_reanchor_on_resume --save_every_sec 1800 --delta_every_steps 0 --delta_every_sec 0 --delta_max_keep 0 --max_ckpts 2 --disk_free_floor_gb 12.0 --heartbeat_every_sec 30 --dblock_fullstack_ar_weight 1.0 --dblock_fullstack_sat_weight 0.1 --dblock_fullstack_nat_weight 0.1 --dblock_local_aux_warn_ce 0.0 --dblock_local_aux_max_ce 0.0 --lr_schedule_origin_tokens 85002387456 --dblock_stop_after_commits 0 --no-oom_auto_backoff --require_exact_resume_state --dblock_objective_mode committed_9_1_1 --dblock_ar_weight 0.25 --dblock_sat_weight 0.25 --dblock_nat_weight 0.25 --dblock_aux_grad_ratio 0.5 --dblock_min_ar_share 0.8 --lr_core 2e-7 --lr_head 2e-7 --batch_size 60 --oom_known_safe_batch 56 --dblock_checkpoint_stride 2 --empty_cache_every_steps 0 --dblock_log_every 25 --cuda_max_reserved_mib 0 --cuda_min_free_mib 0 --weight_accel_mode shadow --weight_accel_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/weight_trajectory_accel --weight_accel_every 1 --weight_accel_train_every 4 --weight_accel_record_every 14 --weight_accel_state_every 256 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2518699",
|
| 425 |
+
"schema": "agillm.launch.v1"
|
| 426 |
+
},
|
| 427 |
+
"local_step": 2539050,
|
| 428 |
+
"loss": 11.288552284240723,
|
| 429 |
+
"nat_mask_contract": "agillm.nat-mask.v1",
|
| 430 |
+
"nat_mask_contract_source": "explicit-legacy-recovery",
|
| 431 |
+
"nat_mask_eos_id": 1,
|
| 432 |
+
"nat_mask_legacy_blank_id": 1,
|
| 433 |
+
"nat_mask_migrated_from_id": 1,
|
| 434 |
+
"nat_mask_schema_version": 1,
|
| 435 |
+
"nat_mask_token_id": 2,
|
| 436 |
+
"pid": 3402475,
|
| 437 |
+
"schema_version": 1,
|
| 438 |
+
"seen_tok": 245714677760,
|
| 439 |
+
"source_is_lane_checkpoint": true,
|
| 440 |
+
"source_is_master_checkpoint": false,
|
| 441 |
+
"source_is_non_master_checkpoint": false,
|
| 442 |
+
"step": 2539050,
|
| 443 |
+
"train_argv": "/workspace/agillm43_production_v14/runtime/agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py train --preset agillm4_floor --block 2048 --source HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25 --nat_mask_token_id 2 --tie_weights --tie_kv --moe_ffn --moe_experts 2 --moe_top_k 1 --moe_mlp_mult 4 --moe_shared_experts 1 --moe_shared_mlp_mult 2 --moe_aux_coef 0.01 --moe_z_coef 0.001 --amp --grad_checkpoint --attn_backend sdpa --sublinear_window 128 --sublinear_stride 128 --sublinear_max_anchors 128 --sublinear_chunk 128 --sublinear_sinks 4 --sublinear_recent_anchors 64 --no-sublinear_pooled_landmarks --anchor_stride 256 --anchor_max 2048 --anchor_position -1 --alibi_mode corrected --alibi_scale 0.0 --optimizer adamw8bit --weight_decay 0.0 --lr_decay cosine --lr_decay_tokens 49000954880 --lr_min_mult 0.2 --lr_schedule_reset_on_resume --lr_warmup_tokens 10000000 --lr_warmup_min_mult 0.1 --dblock --dblock_blocks 14 --dblock_schedule roundrobin --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --dblock_sublayer_mode off --dblock_checkpoint_skip_tail 0 --dblock_loop_layers 0 --dblock_loop_start 0 --dblock_loop_cond_scale 1.0 --dblock_sigma_curriculum_steps 2000 --dblock_sigma_sampling lognormal --dblock_sigma_stratified --dblock_sigma_min 0.002 --dblock_sigma_max 2.0 --dblock_sigma_pmean -1.2 --dblock_sigma_pstd 1.2 --dblock_edm_wmax 5.0 --dblock_ar_prob 0.8 --dblock_sat_prob 0.1 --dblock_nat_prob 0.1 --dblock_ar_loss_tokens 8192 --dblock_sat_loss_tokens 4096 --dblock_nat_loss_tokens 4096 --dblock_nat_embed_noise_mode off --dblock_nat_embed_noise_scale 1.0 --nat_loss_weight 1.0 --nat_mask_ratio 0.5 --nat_max_tokens 0 --nat_span_mask_prob 0.35 --nat_suffix_mask_prob 0.20 --nat_span_max_tokens 0 --loss_spike_skip 0.0 --sat_every 1 --nat_every 1 --dblock_fullstack_anchor_spike_skip 0.0 --dblock_fullstack_anchor_max_ce 0.0 --dblock_anchor_softcap_mult 4.0 --dblock_anchor_softcap_min_ce 25.0 --dblock_anchor_softcap_max_ce 50.0 --dblock_fullstack_ar_every 3 --dblock_fullstack_ar_offset 0 --dblock_fullstack_ar_tokens 256 --dblock_fullstack_sat_every 3 --dblock_fullstack_sat_offset 1 --dblock_fullstack_sat_tokens 128 --dblock_fullstack_nat_every 3 --dblock_fullstack_nat_offset 2 --dblock_fullstack_nat_tokens 128 --dblock_fullstack_nat_mask_id -1 --ckpt_codec block-sharded-zstd --async_update_every_steps 0 --dblock_router_hidden 64 --dblock_router_heads 4 --dblock_router_layers 2 --dblock_router_lr 0.0005 --dblock_explore 0.08 --dblock_max_stale_steps 64 --dblock_max_count_skew 1.35 --dblock_stale_bonus 0.35 --dblock_undertrain_bonus 0.25 --dblock_activation_offload_min_mb 1.0 --nat_document_boundary_aware --dblock_fullstack_anchor_deterministic_eval --dblock_fullstack_anchor_spike_retry_limit 3 --dblock_local_spike_retry_limit 3 --dblock_fullstack_nat_no_valid_crop_limit 8 --target_tokens 400000000000 --lr_schedule_reanchor_on_resume --save_every_sec 1800 --delta_every_steps 0 --delta_every_sec 0 --delta_max_keep 0 --max_ckpts 2 --disk_free_floor_gb 12.0 --heartbeat_every_sec 30 --dblock_fullstack_ar_weight 1.0 --dblock_fullstack_sat_weight 0.1 --dblock_fullstack_nat_weight 0.1 --dblock_local_aux_warn_ce 0.0 --dblock_local_aux_max_ce 0.0 --lr_schedule_origin_tokens 85002387456 --dblock_stop_after_commits 0 --no-oom_auto_backoff --require_exact_resume_state --dblock_objective_mode committed_9_1_1 --dblock_ar_weight 0.25 --dblock_sat_weight 0.25 --dblock_nat_weight 0.25 --dblock_aux_grad_ratio 0.5 --dblock_min_ar_share 0.8 --lr_core 2e-7 --lr_head 2e-7 --batch_size 60 --oom_known_safe_batch 56 --dblock_checkpoint_stride 2 --empty_cache_every_steps 0 --dblock_log_every 25 --cuda_max_reserved_mib 0 --cuda_min_free_mib 0 --weight_accel_mode shadow --weight_accel_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/weight_trajectory_accel --weight_accel_every 1 --weight_accel_train_every 4 --weight_accel_record_every 14 --weight_accel_state_every 256 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2518699",
|
| 444 |
+
"train_script": "agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py",
|
| 445 |
+
"train_script_sha256": "b7690fa26ac7e85f001068e93b0f170baec80d4f40f96241426c2a0582b7e00c",
|
| 446 |
+
"warmstart_base_seen_tok": 0,
|
| 447 |
+
"warmstart_base_step": 2518657,
|
| 448 |
+
"warmstart_kind": "warmstarted_from_lane_checkpoint",
|
| 449 |
+
"warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt",
|
| 450 |
+
"warmstart_source_step": 2518657
|
| 451 |
+
},
|
| 452 |
+
"alibi_mode": "corrected",
|
| 453 |
+
"alibi_scale": 0.0,
|
| 454 |
+
"checkpoint_summary": "warmstarted_from_lane_checkpoint; source=/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt; path=/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 455 |
+
"dataset_provenance": {
|
| 456 |
+
"chat": false,
|
| 457 |
+
"dataset_field_text": "text",
|
| 458 |
+
"dataset_hotload_config_path": "/workspace/agillm43_dataset_hotload.json",
|
| 459 |
+
"dataset_hotload_config_sha256": "2762ba666c280207bdb2293b6d382ef8077d167de79609e5bcc35d4499fd1527",
|
| 460 |
+
"dataset_hotload_config_stamp": [
|
| 461 |
+
73,
|
| 462 |
+
2196426857,
|
| 463 |
+
1787483694198261655,
|
| 464 |
+
1281
|
| 465 |
+
],
|
| 466 |
+
"dataset_hotload_config_used": true,
|
| 467 |
+
"dataset_hotload_enabled": true,
|
| 468 |
+
"dataset_hotload_generation": 4,
|
| 469 |
+
"dataset_hotload_schema": "agillm.dataset_hotload.config.v2",
|
| 470 |
+
"dataset_revision_map_sha256": "8361995a504a2365136b20e1a879ba78944a06309f13f3f013bbd0cdfc72225b",
|
| 471 |
+
"dataset_revisions": {
|
| 472 |
+
"EleutherAI/proof-pile-2": "901a9273a770e9d4138c5ddd91802f9c5c6cdc4b",
|
| 473 |
+
"HuggingFaceFW/fineweb": "9bb295ddab0e05d785b879661af7260fed5140fc",
|
| 474 |
+
"HuggingFaceFW/fineweb-edu": "87f09149ef4734204d70ed1d046ddc9ca3f2b8f9",
|
| 475 |
+
"HuggingFaceTB/cosmopedia": "0ae6ec63f91742bd2d1eaef4f02232c55d719385",
|
| 476 |
+
"HuggingFaceTB/finemath": "e92b25a616738fe95dc186b64dfb19f9c8525594",
|
| 477 |
+
"HuggingFaceTB/smollm-corpus": "3ba9d605774198c5868892d7a8deda78031a781f",
|
| 478 |
+
"Skylion007/openwebtext": "79d93d786212f7344586290adb811d4ae6a1762c",
|
| 479 |
+
"allenai/c4": "1588ec454efa1a09f29cd18ddd04fe05fc8653a2",
|
| 480 |
+
"allenai/dolma": "7f48140530a023e9ea4c5cfb141160922727d4d3",
|
| 481 |
+
"allenai/dolmino-mix-1124": "a319f19eef1e257417b11ea8c30da266ae175557",
|
| 482 |
+
"codeparrot/codeparrot-clean": "35a59fb025bc0a102f7d96eac09d145b896d487b",
|
| 483 |
+
"mlfoundations/dclm-baseline-1.0": "a3b142c183aebe5af344955ae20836eb34dcf69b",
|
| 484 |
+
"wikimedia/wikipedia": "b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
|
| 485 |
+
},
|
| 486 |
+
"dolma_payload_integrity": "external_url_manifest_pinned_payload_hashes_unavailable",
|
| 487 |
+
"dolma_url_list_sha256": "bc6ef505be920c94bb9e7147445638f606fdc3f5af622cd1d887dc48e21a2d5b",
|
| 488 |
+
"dolma_url_manifest_raw_sha256": "6fe8c40d51e5bebc9303453d7f533b232d69f7ae122123621a82621b1cd779b7",
|
| 489 |
+
"has_language_mix": true,
|
| 490 |
+
"has_numeracy": false,
|
| 491 |
+
"hot_config_expected_sha256": "",
|
| 492 |
+
"hot_config_has_append": true,
|
| 493 |
+
"hot_config_has_datasets": false,
|
| 494 |
+
"hot_config_mtime": 1787722590.529335,
|
| 495 |
+
"hot_config_path": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
|
| 496 |
+
"hot_config_used": true,
|
| 497 |
+
"hot_reload_capable": true,
|
| 498 |
+
"phase": "pretrain",
|
| 499 |
+
"proof_pile_lfs_manifest_sha256": "d020dfb796b02a75564e4e11d9e03185a55b3f9c6639f87830cfbe10737ed329",
|
| 500 |
+
"proof_pile_paths_sha256": "8a30018017e3b3d9a9bcfa06a3c404ccd113297b0b1d0c48e2e2aadf113e52b0",
|
| 501 |
+
"schema": "agillm.dataset_provenance.v1",
|
| 502 |
+
"sequence_prefetch_depth": 256,
|
| 503 |
+
"source_count": 15,
|
| 504 |
+
"source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25,HuggingFaceTB/finemath:finemath-4plus|0.7,HuggingFaceTB/finemath:infiwebmath-4plus|0.35,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8,allenai/dolmino-mix-1124|1,mlfoundations/dclm-baseline-1.0|1,HuggingFaceFW/fineweb-edu:sample-100BT|1.6",
|
| 505 |
+
"source_requested": "HuggingFaceFW/fineweb-edu:sample-10BT|1.60,HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25",
|
| 506 |
+
"source_specs": [
|
| 507 |
+
"HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50",
|
| 508 |
+
"wikimedia/wikipedia:20231101.en|1.10",
|
| 509 |
+
"allenai/c4:en|0.40",
|
| 510 |
+
"Skylion007/openwebtext|0.35",
|
| 511 |
+
"HuggingFaceTB/cosmopedia:web_samples_v2|1.30",
|
| 512 |
+
"HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30",
|
| 513 |
+
"EleutherAI/proof-pile-2:all|1.50",
|
| 514 |
+
"allenai/dolma:v1_6-sample|0.50",
|
| 515 |
+
"codeparrot/codeparrot-clean|1.25",
|
| 516 |
+
"HuggingFaceTB/finemath:finemath-4plus|0.7",
|
| 517 |
+
"HuggingFaceTB/finemath:infiwebmath-4plus|0.35",
|
| 518 |
+
"HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8",
|
| 519 |
+
"allenai/dolmino-mix-1124|1",
|
| 520 |
+
"mlfoundations/dclm-baseline-1.0|1",
|
| 521 |
+
"HuggingFaceFW/fineweb-edu:sample-100BT|1.6"
|
| 522 |
+
],
|
| 523 |
+
"val_source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25,HuggingFaceTB/finemath:finemath-4plus|0.7,HuggingFaceTB/finemath:infiwebmath-4plus|0.35,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8,allenai/dolmino-mix-1124|1,mlfoundations/dclm-baseline-1.0|1,HuggingFaceFW/fineweb-edu:sample-100BT|1.6",
|
| 524 |
+
"val_source_requested": ""
|
| 525 |
+
},
|
| 526 |
+
"lr_schedule_origin_tok": 85002387456,
|
| 527 |
+
"nat_mask_contract": "agillm.nat-mask.v1",
|
| 528 |
+
"nat_mask_contract_source": "explicit-legacy-recovery",
|
| 529 |
+
"nat_mask_eos_id": 1,
|
| 530 |
+
"nat_mask_legacy_blank_id": 1,
|
| 531 |
+
"nat_mask_migrated_from_id": 1,
|
| 532 |
+
"nat_mask_schema_version": 1,
|
| 533 |
+
"nat_mask_token_id": 2,
|
| 534 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 535 |
+
"promotion_required": true,
|
| 536 |
+
"quality_gate_used_for_training": false,
|
| 537 |
+
"repair_schema": "",
|
| 538 |
+
"saved_at": "2026-08-30T08:54:18Z",
|
| 539 |
+
"schema": "agillm43.serving.candidate.v1",
|
| 540 |
+
"source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.50,wikimedia/wikipedia:20231101.en|1.10,allenai/c4:en|0.40,Skylion007/openwebtext|0.35,HuggingFaceTB/cosmopedia:web_samples_v2|1.30,HuggingFaceTB/smollm-corpus:cosmopedia-v2|1.30,EleutherAI/proof-pile-2:all|1.50,allenai/dolma:v1_6-sample|0.50,codeparrot/codeparrot-clean|1.25,HuggingFaceTB/finemath:finemath-4plus|0.7,HuggingFaceTB/finemath:infiwebmath-4plus|0.35,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|0.8,allenai/dolmino-mix-1124|1,mlfoundations/dclm-baseline-1.0|1,HuggingFaceFW/fineweb-edu:sample-100BT|1.6",
|
| 541 |
+
"step": 2539050,
|
| 542 |
+
"training_pointer_schema": "agillm43.training.resume.v2",
|
| 543 |
+
"training_pointer_sha256": "b561322fa5dbdf3e94b257e2e7d537b93842e2c4dff320ce94a029319c5b72bf",
|
| 544 |
+
"warmstart_kind": "warmstarted_from_lane_checkpoint",
|
| 545 |
+
"warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt"
|
| 546 |
+
}
|
checkpoints/step2539050_20260830/training_latest.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint_main_nbytes": 17669584,
|
| 3 |
+
"checkpoint_main_sha256": "8111b494894e7a87d8fb2661cdabb074c7654f0221b4b93a305e308cb25fc4f8",
|
| 4 |
+
"checkpoint_name": "pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 5 |
+
"checkpoint_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 6 |
+
"continuation_state_schema": "agillm43.continuation.state.v2",
|
| 7 |
+
"continuation_state_sha256": "920effdaf62e57c72723f3d5a3ced083f1c5f1bff8b8d14f16c934c9f1ebaa3f",
|
| 8 |
+
"dblock_committed_step": 808699,
|
| 9 |
+
"package_format": "block-sharded",
|
| 10 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt",
|
| 11 |
+
"quality_gate_used_for_training": false,
|
| 12 |
+
"record_sha256": "b561322fa5dbdf3e94b257e2e7d537b93842e2c4dff320ce94a029319c5b72bf",
|
| 13 |
+
"role": "training",
|
| 14 |
+
"saved_at": "2026-08-30T08:54:18Z",
|
| 15 |
+
"schema": "agillm43.training.resume.v2",
|
| 16 |
+
"seen_tok": 245714677760,
|
| 17 |
+
"shard_dir": "pretrain_step02539050_from02518657_20260830T0853Z.pt.shards",
|
| 18 |
+
"shards": [
|
| 19 |
+
{
|
| 20 |
+
"name": "core_base.pt",
|
| 21 |
+
"nbytes": 608343998,
|
| 22 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_base.pt",
|
| 23 |
+
"sha256": "b66dffd15458d0c732b5372cc31fa39e8705688610ba88b2cc798b5889991f27"
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"name": "core_block_0000.pt",
|
| 27 |
+
"nbytes": 117685010,
|
| 28 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0000.pt",
|
| 29 |
+
"sha256": "6541734b0e2fc4f214a29f41bda15b68af23bb366584540864fe43398e0a47e6"
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"name": "core_block_0001.pt",
|
| 33 |
+
"nbytes": 116969620,
|
| 34 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0001.pt",
|
| 35 |
+
"sha256": "a801faffdf915a0829a97437fe4db1dcbcd50ac2d12cafb9126202a90d7e1cf1"
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"name": "core_block_0002.pt",
|
| 39 |
+
"nbytes": 116949506,
|
| 40 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0002.pt",
|
| 41 |
+
"sha256": "6d6805e384e5033a357991958bb435d0c8dc4343d77d51948dce63ad1ddc68a4"
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"name": "core_block_0003.pt",
|
| 45 |
+
"nbytes": 116926589,
|
| 46 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0003.pt",
|
| 47 |
+
"sha256": "d51fde4a186d8851348533dd76a1a1996ad6d037934c6252cfc8f259aa2478d9"
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"name": "core_block_0004.pt",
|
| 51 |
+
"nbytes": 117018980,
|
| 52 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0004.pt",
|
| 53 |
+
"sha256": "537f19a8e40dc5bc7f5eb3fc070eccfb95f2a4cfc0b6b003c63aca82f8c47374"
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"name": "core_block_0005.pt",
|
| 57 |
+
"nbytes": 116981931,
|
| 58 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0005.pt",
|
| 59 |
+
"sha256": "f128cf8326e1dada31e0c18c546b668401d72290a6e6c5054bd1c9631cfa7bc2"
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"name": "core_block_0006.pt",
|
| 63 |
+
"nbytes": 117365872,
|
| 64 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0006.pt",
|
| 65 |
+
"sha256": "57b58fc9c155ba4e629cd04b8fc3c3bcc4af1a445357872ba1501f4669249cb6"
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"name": "core_block_0007.pt",
|
| 69 |
+
"nbytes": 117604270,
|
| 70 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0007.pt",
|
| 71 |
+
"sha256": "cde867e856d8eb35fb56faf4aadeec532532f89be9645d0fd63bcc664d53ade7"
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"name": "core_block_0008.pt",
|
| 75 |
+
"nbytes": 117454437,
|
| 76 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0008.pt",
|
| 77 |
+
"sha256": "01a92db08fe30b3b3758b928a32a854c117e26ab9f175ec88cf3083db78f4f36"
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"name": "core_block_0009.pt",
|
| 81 |
+
"nbytes": 117280515,
|
| 82 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0009.pt",
|
| 83 |
+
"sha256": "876aba45959990b1b649fa97e171a1ce617588e63d218e605d606581ba4e1257"
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"name": "core_block_0010.pt",
|
| 87 |
+
"nbytes": 117411716,
|
| 88 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0010.pt",
|
| 89 |
+
"sha256": "f9daaa77e43bd3d2194fe18d26781c1401956d5417e24f29c35aa619a7ed8cb1"
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"name": "core_block_0011.pt",
|
| 93 |
+
"nbytes": 117420204,
|
| 94 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0011.pt",
|
| 95 |
+
"sha256": "b4847a30f910ea7c54f2764448c12b4f01ea71fb8b9f5d390eace79557530258"
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"name": "core_block_0012.pt",
|
| 99 |
+
"nbytes": 123847916,
|
| 100 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0012.pt",
|
| 101 |
+
"sha256": "f18452f5f09b9b190c4e6148f302f263b02ce6f1fabae95ea46cf9bb55331a90"
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"name": "core_block_0013.pt",
|
| 105 |
+
"nbytes": 120977605,
|
| 106 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0013.pt",
|
| 107 |
+
"sha256": "37f5e584921826f698cf00cd8b056aee756b6f4e690d89f347a76500fc836f7c"
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"name": "core_block_0014.pt",
|
| 111 |
+
"nbytes": 117789828,
|
| 112 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0014.pt",
|
| 113 |
+
"sha256": "631986c4759a30b829b10d458e9b438378e938d7562b68f813be198d541295a5"
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"name": "core_block_0015.pt",
|
| 117 |
+
"nbytes": 116988336,
|
| 118 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0015.pt",
|
| 119 |
+
"sha256": "07461518d666903a0bbdd4524a5ad8488f8002473a5d6997361f2823eb9005ec"
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"name": "core_block_0016.pt",
|
| 123 |
+
"nbytes": 117233272,
|
| 124 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0016.pt",
|
| 125 |
+
"sha256": "69018d28a94c93687c422e981ae5414251a5a2aa7a545503b18ebfc8c611aee6"
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"name": "core_block_0017.pt",
|
| 129 |
+
"nbytes": 116966011,
|
| 130 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0017.pt",
|
| 131 |
+
"sha256": "2688dc7732e5dd7959ccf12adbae2dba14ddd92b2b4c74c8fd0207977606612e"
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"name": "core_block_0018.pt",
|
| 135 |
+
"nbytes": 117109412,
|
| 136 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0018.pt",
|
| 137 |
+
"sha256": "a2818981dda6b122a24cce1cc074fc0f6aa291e452ba02a6eb4219097a7d3ab8"
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"name": "core_block_0019.pt",
|
| 141 |
+
"nbytes": 116995972,
|
| 142 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0019.pt",
|
| 143 |
+
"sha256": "592f15ad2347415dcaaf545f7181121d2b0c765bc5833c84f5e10005ed05c978"
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"name": "core_block_0020.pt",
|
| 147 |
+
"nbytes": 117028154,
|
| 148 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0020.pt",
|
| 149 |
+
"sha256": "d72d7bd2422fc6ea85ece7b8e893dc4f6672eef328a78f6296737d9d7879fd8e"
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"name": "core_block_0021.pt",
|
| 153 |
+
"nbytes": 116872053,
|
| 154 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0021.pt",
|
| 155 |
+
"sha256": "c39162f97413cc51fa0180c5cbc507ad78f49fa2cb25d933c3eceb38d173e7d6"
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"name": "core_block_0022.pt",
|
| 159 |
+
"nbytes": 117046446,
|
| 160 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0022.pt",
|
| 161 |
+
"sha256": "ef822aa36029d5033fd505e32cbe5fbbfcf5d623df747c28d565ad5cb15daae6"
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"name": "core_block_0023.pt",
|
| 165 |
+
"nbytes": 117318411,
|
| 166 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0023.pt",
|
| 167 |
+
"sha256": "8d5672b39d8c54b2be671a3e3864a56144cf8842d8df431cca4a9fe3bde3625b"
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"name": "core_block_0024.pt",
|
| 171 |
+
"nbytes": 121465345,
|
| 172 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0024.pt",
|
| 173 |
+
"sha256": "8397e970e86f7f0c52c40928b32f7841033513961b877b96c208c149debaf4e0"
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"name": "core_block_0025.pt",
|
| 177 |
+
"nbytes": 121496088,
|
| 178 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0025.pt",
|
| 179 |
+
"sha256": "556366401ef13b51b9f28668e3d976cb4ae5f712b07b2503fc6157d79962068e"
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"name": "core_block_0026.pt",
|
| 183 |
+
"nbytes": 121424063,
|
| 184 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0026.pt",
|
| 185 |
+
"sha256": "d2d44504f5e25c8967410eecb3ccf1d46afc9c6534aa7b914305f08311691c91"
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"name": "core_block_0027.pt",
|
| 189 |
+
"nbytes": 121617002,
|
| 190 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/core_block_0027.pt",
|
| 191 |
+
"sha256": "16f23b6102584128fb33b7b87b906a0d62cef9a34dd450c2c884e71fe925a096"
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"name": "ar.pt",
|
| 195 |
+
"nbytes": 608347130,
|
| 196 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/ar.pt",
|
| 197 |
+
"sha256": "8ac2ffec05fb48a7765ced76fc6ea988fc6793c879925e402e2f572acff0d395"
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"name": "sat.pt",
|
| 201 |
+
"nbytes": 608342126,
|
| 202 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/sat.pt",
|
| 203 |
+
"sha256": "f1dbaaa21a94b07666ec9114b5b270dd232b5a34991378a5ff880db8d3d0146b"
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"name": "nat.pt",
|
| 207 |
+
"nbytes": 608347130,
|
| 208 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/nat.pt",
|
| 209 |
+
"sha256": "8ac2ffec05fb48a7765ced76fc6ea988fc6793c879925e402e2f572acff0d395"
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"name": "opt.pt",
|
| 213 |
+
"nbytes": 1577355158,
|
| 214 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/opt.pt",
|
| 215 |
+
"sha256": "5a9e0fed590c82ab2f15f19952b4ccf9d9f340abb00cfb94e279f6a8224f37e1"
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"name": "scaler.pt",
|
| 219 |
+
"nbytes": 149,
|
| 220 |
+
"path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.shards/scaler.pt",
|
| 221 |
+
"sha256": "c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c"
|
| 222 |
+
}
|
| 223 |
+
],
|
| 224 |
+
"step": 2539050,
|
| 225 |
+
"tokenizer_sidecar_nbytes": 17465482,
|
| 226 |
+
"tokenizer_sidecar_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02539050_from02518657_20260830T0853Z.pt.tokenizer.json",
|
| 227 |
+
"tokenizer_sidecar_sha256": "65378a6b6b6b3b5686a24123581b2451937947fca125230210366771242db3da"
|
| 228 |
+
}
|