Hugging Face's logo

MarxistLeninist
/
AGILLM-4.3-checkpoints

Model card Files
xet
 Community
417
MarxistLeninist commited on
Commit
19cbda2
·
1 Parent(s): 0e9517d

checkpoint step2801622 (#378)

Browse files

- 1ef4e88268e6a85eadd3fde53f00c5046f4c1b616e081f560024ba73b821ac5b (f43d4d6c9026f4b3fdf00d36ad1794d9e924a4a4)
- dcf4c00a8fd6377e05b9515ed980a8a3cddf8faafed77e62301f3340ebdde106 (13ab676131e19667013901c934b930b30a1bf6d6)
- be322370c814a0977ed01fb140dc0c84c9b70e3b3bc007c75c81efe74174ea5c (ffd7fc823f9c4fb88974fe2c847a6acf589b7e77)
- 576ca99982cb37344d5e25744a6bc93b39d8532ca9be7747d14acb4f6265ee8e (0bd3cbc46ba99c9ce73b5a5730ca29d40a9fc13e)
- b5863a549fde60a0ada1921d6668de97f19315c846e65a60e62378a2fd1d75ab (27300b9f0b9c19e88917d566d3bdbb4a10488e79)

Files changed (5) hide show
  1. .gitattributes +1 -0
  2. checkpoints/step2801622_20260908/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_scaler.pt +3 -0
  3. checkpoints/step2801622_20260908/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/scaler.pt +3 -0
  4. checkpoints/step2801622_20260908/serving_candidate.json +679 -0
  5. checkpoints/step2801622_20260908/training_latest.json +455 -0
.gitattributes CHANGED
@@ -396,3 +396,4 @@ checkpoints/step2794518_20260907/pretrain_step02794518_from02787566_20260907T112
396
  checkpoints/step2795579_20260907/pretrain_step02795579_from02787566_20260907T140616_1788789976395929304_3798343_000061.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
397
  checkpoints/step2796124_20260907/pretrain_step02796124_from02787566_20260907T152617_1788794777175265822_3798343_000065.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
398
  checkpoints/step2796987_20260907/pretrain_step02796987_from02787566_20260907T173520_1788802520608359093_3798343_000072.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
396
  checkpoints/step2795579_20260907/pretrain_step02795579_from02787566_20260907T140616_1788789976395929304_3798343_000061.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
397
  checkpoints/step2796124_20260907/pretrain_step02796124_from02787566_20260907T152617_1788794777175265822_3798343_000065.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
398
  checkpoints/step2796987_20260907/pretrain_step02796987_from02787566_20260907T173520_1788802520608359093_3798343_000072.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
399
+ checkpoints/step2801622_20260908/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
checkpoints/step2801622_20260908/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c
3
+ size 149
checkpoints/step2801622_20260908/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.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/step2801622_20260908/serving_candidate.json ADDED
@@ -0,0 +1,679 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
6
+ "checkpoint_type": "full",
7
+ "created_at_iso": "2026-09-08T05:30:07Z",
8
+ "created_at_unix": 1788845407.5713515,
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": "61b8f59b84791d4dbee2a72a2ff4d0c9837f930f95cf94ab0013daf2abf1ae51",
15
+ "dataset_hotload_config_stamp": [
16
+ 50,
17
+ 1119951817,
18
+ 1788666957977200004,
19
+ 4227
20
+ ],
21
+ "dataset_hotload_config_used": true,
22
+ "dataset_hotload_enabled": true,
23
+ "dataset_hotload_generation": 10,
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": 1788708887.095225,
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": 16,
59
+ "source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96,wikimedia/wikipedia:20231101.en|1.63,allenai/c4:en|0.36,Skylion007/openwebtext|0.18,HuggingFaceTB/cosmopedia:web_samples_v2|0.64,HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65,EleutherAI/proof-pile-2:all|0.67,allenai/dolma:v1_6-sample|0.5,codeparrot/codeparrot-clean|0.24,HuggingFaceTB/finemath:finemath-4plus|0.32,HuggingFaceTB/finemath:infiwebmath-4plus|0.23,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05,allenai/dolmino-mix-1124|0.78,mlfoundations/dclm-baseline-1.0|1.05,HuggingFaceFW/fineweb-edu:sample-100BT|2.4,json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03",
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|0.96",
63
+ "wikimedia/wikipedia:20231101.en|1.63",
64
+ "allenai/c4:en|0.36",
65
+ "Skylion007/openwebtext|0.18",
66
+ "HuggingFaceTB/cosmopedia:web_samples_v2|0.64",
67
+ "HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65",
68
+ "EleutherAI/proof-pile-2:all|0.67",
69
+ "allenai/dolma:v1_6-sample|0.5",
70
+ "codeparrot/codeparrot-clean|0.24",
71
+ "HuggingFaceTB/finemath:finemath-4plus|0.32",
72
+ "HuggingFaceTB/finemath:infiwebmath-4plus|0.23",
73
+ "HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05",
74
+ "allenai/dolmino-mix-1124|0.78",
75
+ "mlfoundations/dclm-baseline-1.0|1.05",
76
+ "HuggingFaceFW/fineweb-edu:sample-100BT|2.4",
77
+ "json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03"
78
+ ],
79
+ "val_source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96,wikimedia/wikipedia:20231101.en|1.63,allenai/c4:en|0.36,Skylion007/openwebtext|0.18,HuggingFaceTB/cosmopedia:web_samples_v2|0.64,HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65,EleutherAI/proof-pile-2:all|0.67,allenai/dolma:v1_6-sample|0.5,codeparrot/codeparrot-clean|0.24,HuggingFaceTB/finemath:finemath-4plus|0.32,HuggingFaceTB/finemath:infiwebmath-4plus|0.23,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05,allenai/dolmino-mix-1124|0.78,mlfoundations/dclm-baseline-1.0|1.05,HuggingFaceFW/fineweb-edu:sample-100BT|2.4,json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03",
80
+ "val_source_requested": ""
81
+ },
82
+ "effective_global_step": 5589188,
83
+ "effective_seen_tok": 277974865920,
84
+ "global_origin_seen_tok": 0,
85
+ "global_origin_step": 2787566,
86
+ "gpu": {
87
+ "allocated_gb": 7.13,
88
+ "peak_allocated_gb": 15.08,
89
+ "reserved_gb": 17.34
90
+ },
91
+ "hostname": "ead6505f92c4",
92
+ "lane": "pretrain",
93
+ "launch": {
94
+ "argv": [
95
+ "/workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906/agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.py",
96
+ "train",
97
+ "--preset",
98
+ "agillm4_floor",
99
+ "--block",
100
+ "2048",
101
+ "--source",
102
+ "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",
103
+ "--nat_mask_token_id",
104
+ "2",
105
+ "--tie_weights",
106
+ "--tie_kv",
107
+ "--moe_ffn",
108
+ "--moe_experts",
109
+ "2",
110
+ "--moe_top_k",
111
+ "1",
112
+ "--moe_mlp_mult",
113
+ "4",
114
+ "--moe_shared_experts",
115
+ "1",
116
+ "--moe_shared_mlp_mult",
117
+ "2",
118
+ "--moe_aux_coef",
119
+ "0.01",
120
+ "--moe_z_coef",
121
+ "0.001",
122
+ "--amp",
123
+ "--grad_checkpoint",
124
+ "--attn_backend",
125
+ "sdpa",
126
+ "--sublinear_window",
127
+ "128",
128
+ "--sublinear_stride",
129
+ "128",
130
+ "--sublinear_max_anchors",
131
+ "128",
132
+ "--sublinear_chunk",
133
+ "128",
134
+ "--sublinear_sinks",
135
+ "4",
136
+ "--sublinear_recent_anchors",
137
+ "64",
138
+ "--no-sublinear_pooled_landmarks",
139
+ "--anchor_stride",
140
+ "256",
141
+ "--anchor_max",
142
+ "2048",
143
+ "--anchor_position",
144
+ "-1",
145
+ "--alibi_mode",
146
+ "corrected",
147
+ "--alibi_scale",
148
+ "0.0",
149
+ "--optimizer",
150
+ "adamw8bit",
151
+ "--weight_decay",
152
+ "0.0",
153
+ "--lr_decay",
154
+ "cosine",
155
+ "--lr_decay_tokens",
156
+ "49000954880",
157
+ "--lr_min_mult",
158
+ "0.2",
159
+ "--lr_schedule_reset_on_resume",
160
+ "--lr_warmup_tokens",
161
+ "10000000",
162
+ "--lr_warmup_min_mult",
163
+ "0.1",
164
+ "--dblock",
165
+ "--dblock_schedule",
166
+ "roundrobin",
167
+ "--dblock_router_ramp_steps",
168
+ "64",
169
+ "--dblock_warmup_steps",
170
+ "14",
171
+ "--dblock_checkpoint_skip_tail",
172
+ "0",
173
+ "--dblock_loop_layers",
174
+ "0",
175
+ "--dblock_loop_start",
176
+ "0",
177
+ "--dblock_loop_cond_scale",
178
+ "1.0",
179
+ "--dblock_sigma_curriculum_steps",
180
+ "2000",
181
+ "--dblock_sigma_sampling",
182
+ "lognormal",
183
+ "--dblock_sigma_stratified",
184
+ "--dblock_sigma_min",
185
+ "0.002",
186
+ "--dblock_sigma_max",
187
+ "2.0",
188
+ "--dblock_sigma_pmean",
189
+ "-1.2",
190
+ "--dblock_sigma_pstd",
191
+ "1.2",
192
+ "--dblock_edm_wmax",
193
+ "5.0",
194
+ "--dblock_ar_prob",
195
+ "0.8",
196
+ "--dblock_sat_prob",
197
+ "0.1",
198
+ "--dblock_nat_prob",
199
+ "0.1",
200
+ "--dblock_ar_loss_tokens",
201
+ "8192",
202
+ "--dblock_sat_loss_tokens",
203
+ "4096",
204
+ "--dblock_nat_loss_tokens",
205
+ "4096",
206
+ "--dblock_nat_embed_noise_mode",
207
+ "off",
208
+ "--dblock_nat_embed_noise_scale",
209
+ "1.0",
210
+ "--nat_loss_weight",
211
+ "1.0",
212
+ "--nat_mask_ratio",
213
+ "0.5",
214
+ "--nat_max_tokens",
215
+ "0",
216
+ "--nat_span_mask_prob",
217
+ "0.35",
218
+ "--nat_suffix_mask_prob",
219
+ "0.20",
220
+ "--nat_span_max_tokens",
221
+ "0",
222
+ "--loss_spike_skip",
223
+ "0.0",
224
+ "--sat_every",
225
+ "1",
226
+ "--nat_every",
227
+ "1",
228
+ "--dblock_fullstack_anchor_spike_skip",
229
+ "0.0",
230
+ "--dblock_fullstack_anchor_max_ce",
231
+ "0.0",
232
+ "--dblock_anchor_softcap_mult",
233
+ "4.0",
234
+ "--dblock_anchor_softcap_min_ce",
235
+ "25.0",
236
+ "--dblock_anchor_softcap_max_ce",
237
+ "50.0",
238
+ "--dblock_fullstack_ar_every",
239
+ "3",
240
+ "--dblock_fullstack_ar_offset",
241
+ "0",
242
+ "--dblock_fullstack_ar_tokens",
243
+ "256",
244
+ "--dblock_fullstack_sat_every",
245
+ "3",
246
+ "--dblock_fullstack_sat_offset",
247
+ "1",
248
+ "--dblock_fullstack_sat_tokens",
249
+ "128",
250
+ "--dblock_fullstack_nat_every",
251
+ "3",
252
+ "--dblock_fullstack_nat_offset",
253
+ "2",
254
+ "--dblock_fullstack_nat_tokens",
255
+ "128",
256
+ "--dblock_fullstack_nat_mask_id",
257
+ "-1",
258
+ "--ckpt_codec",
259
+ "block-sharded-zstd",
260
+ "--async_update_every_steps",
261
+ "0",
262
+ "--dblock_router_hidden",
263
+ "64",
264
+ "--dblock_router_heads",
265
+ "4",
266
+ "--dblock_router_layers",
267
+ "2",
268
+ "--dblock_router_lr",
269
+ "0.0005",
270
+ "--dblock_explore",
271
+ "0.08",
272
+ "--dblock_max_stale_steps",
273
+ "64",
274
+ "--dblock_max_count_skew",
275
+ "1.35",
276
+ "--dblock_stale_bonus",
277
+ "0.35",
278
+ "--dblock_undertrain_bonus",
279
+ "0.25",
280
+ "--dblock_activation_offload_min_mb",
281
+ "1.0",
282
+ "--nat_document_boundary_aware",
283
+ "--dblock_fullstack_anchor_deterministic_eval",
284
+ "--dblock_fullstack_anchor_spike_retry_limit",
285
+ "3",
286
+ "--dblock_local_spike_retry_limit",
287
+ "3",
288
+ "--dblock_fullstack_nat_no_valid_crop_limit",
289
+ "8",
290
+ "--target_tokens",
291
+ "400000000000",
292
+ "--lr_schedule_reanchor_on_resume",
293
+ "--save_every_sec",
294
+ "1800",
295
+ "--delta_max_keep",
296
+ "0",
297
+ "--max_ckpts",
298
+ "2",
299
+ "--disk_free_floor_gb",
300
+ "12.0",
301
+ "--heartbeat_every_sec",
302
+ "30",
303
+ "--dblock_fullstack_ar_weight",
304
+ "1.0",
305
+ "--dblock_fullstack_sat_weight",
306
+ "0.1",
307
+ "--dblock_fullstack_nat_weight",
308
+ "0.1",
309
+ "--dblock_local_aux_warn_ce",
310
+ "0.0",
311
+ "--dblock_local_aux_max_ce",
312
+ "0.0",
313
+ "--lr_schedule_origin_tokens",
314
+ "85002387456",
315
+ "--dblock_stop_after_commits",
316
+ "0",
317
+ "--no-oom_auto_backoff",
318
+ "--require_exact_resume_state",
319
+ "--dblock_objective_mode",
320
+ "committed_9_1_1",
321
+ "--dblock_ar_weight",
322
+ "0.25",
323
+ "--dblock_sat_weight",
324
+ "0.25",
325
+ "--dblock_nat_weight",
326
+ "0.25",
327
+ "--dblock_aux_grad_ratio",
328
+ "0.5",
329
+ "--dblock_min_ar_share",
330
+ "0.8",
331
+ "--lr_core",
332
+ "2e-7",
333
+ "--lr_head",
334
+ "2e-7",
335
+ "--oom_known_safe_batch",
336
+ "56",
337
+ "--dblock_checkpoint_stride",
338
+ "2",
339
+ "--empty_cache_every_steps",
340
+ "0",
341
+ "--dblock_log_every",
342
+ "25",
343
+ "--cuda_max_reserved_mib",
344
+ "0",
345
+ "--cuda_min_free_mib",
346
+ "0",
347
+ "--dblock_clean_context",
348
+ "e2e",
349
+ "--dblock_local_rows",
350
+ "2",
351
+ "--dblock_clean_context_noise_scale",
352
+ "0.0",
353
+ "--dblock_depth_anchor_every",
354
+ "50",
355
+ "--dblock_fullstack_anchor_cap",
356
+ "16376",
357
+ "--dblock_fullstack_ar_rows",
358
+ "8",
359
+ "--detachable_50m_upgrade_multimode",
360
+ "--detachable_50m_upgrade_logit_distill",
361
+ "--detachable_50m_upgrade_sat_gate_v2713",
362
+ "--detachable_50m_create_from_legacy",
363
+ "--detachable_50m_seed_from_mature",
364
+ "--detachable_50m_lr",
365
+ "0.0001",
366
+ "--detachable_50m_weight_decay",
367
+ "0.0",
368
+ "--detachable_50m_update_mode",
369
+ "sublayer",
370
+ "--detachable_50m_batch",
371
+ "1",
372
+ "--detachable_50m_tokens",
373
+ "64",
374
+ "--detachable_50m_loss_tokens",
375
+ "64",
376
+ "--detachable_50m_every",
377
+ "1",
378
+ "--detachable_50m_teacher_every",
379
+ "1",
380
+ "--detachable_50m_distill_weight",
381
+ "1.0",
382
+ "--detachable_50m_logit_distill_alpha",
383
+ "0.30",
384
+ "--detachable_50m_logit_distill_temperature",
385
+ "1.0",
386
+ "--detachable_50m_logit_distill_tokens",
387
+ "64",
388
+ "--detachable_50m_hidden_distill_weight",
389
+ "0.0",
390
+ "--detachable_50m_anchor_every",
391
+ "128",
392
+ "--detachable_50m_anchor_tokens",
393
+ "32",
394
+ "--detachable_50m_seed",
395
+ "20260830",
396
+ "--detachable_50m_log_every",
397
+ "10",
398
+ "--detachable_50m_telemetry_keep",
399
+ "128",
400
+ "--weight_accel_mode",
401
+ "shadow",
402
+ "--weight_accel_dir",
403
+ "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/weight_trajectory_accel",
404
+ "--weight_accel_every",
405
+ "1",
406
+ "--weight_accel_train_every",
407
+ "4",
408
+ "--weight_accel_record_every",
409
+ "14",
410
+ "--weight_accel_state_every",
411
+ "256",
412
+ "--detachable_50m",
413
+ "--detachable_50m_optimizer",
414
+ "paged_adamw8bit",
415
+ "--delta_every_steps",
416
+ "0",
417
+ "--delta_every_sec",
418
+ "0",
419
+ "--resume",
420
+ "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt",
421
+ "--save_dir",
422
+ "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B",
423
+ "--data_seed",
424
+ "2787608",
425
+ "--batch_size",
426
+ "60",
427
+ "--dblock_blocks",
428
+ "14",
429
+ "--dblock_direct56_enabled",
430
+ "1",
431
+ "--dblock_layers_per_update",
432
+ "0",
433
+ "--dblock_router",
434
+ "shadow",
435
+ "--dblock_router_blend",
436
+ "0.0",
437
+ "--dblock_router_full_anchor_every",
438
+ "128",
439
+ "--dblock_router_train_layer_policy",
440
+ "cyclic",
441
+ "--dblock_router_train_layers_per_update",
442
+ "1",
443
+ "--dblock_sublayer_mode",
444
+ "off",
445
+ "--dblock_target_route_activate_from_checkpoint",
446
+ "1",
447
+ "--dblock_target_route_activation_checkpoint_step",
448
+ "2717509",
449
+ "--dblock_target_route_activation_parent_pointer_sha256",
450
+ "04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827",
451
+ "--dblock_target_route_offset",
452
+ "1",
453
+ "--dblock_target_route_policy",
454
+ "xor_fold56_v1",
455
+ "--dblock_train_layer_policy",
456
+ "cyclic",
457
+ "--dblock_train_layers_per_update",
458
+ "1",
459
+ "--dblock_train_sublayer_policy",
460
+ "cyclic",
461
+ "--dblock_train_sublayers_per_update",
462
+ "1",
463
+ "--granularity_label",
464
+ "one-sublayer-local-grad-direct56-teacher-v27.16",
465
+ "--val_every_sec",
466
+ "600",
467
+ "--val_file",
468
+ "/workspace/agillm43_heldout_v3/token_ids.json",
469
+ "--val_sha256",
470
+ "513b30203ffd18167655deb2d9cd610541df87d1ea0b8a0f9241c58e68703d21"
471
+ ],
472
+ "argv_string": "/workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906/agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.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_schedule roundrobin --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --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_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 --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 --dblock_clean_context e2e --dblock_local_rows 2 --dblock_clean_context_noise_scale 0.0 --dblock_depth_anchor_every 50 --dblock_fullstack_anchor_cap 16376 --dblock_fullstack_ar_rows 8 --detachable_50m_upgrade_multimode --detachable_50m_upgrade_logit_distill --detachable_50m_upgrade_sat_gate_v2713 --detachable_50m_create_from_legacy --detachable_50m_seed_from_mature --detachable_50m_lr 0.0001 --detachable_50m_weight_decay 0.0 --detachable_50m_update_mode sublayer --detachable_50m_batch 1 --detachable_50m_tokens 64 --detachable_50m_loss_tokens 64 --detachable_50m_every 1 --detachable_50m_teacher_every 1 --detachable_50m_distill_weight 1.0 --detachable_50m_logit_distill_alpha 0.30 --detachable_50m_logit_distill_temperature 1.0 --detachable_50m_logit_distill_tokens 64 --detachable_50m_hidden_distill_weight 0.0 --detachable_50m_anchor_every 128 --detachable_50m_anchor_tokens 32 --detachable_50m_seed 20260830 --detachable_50m_log_every 10 --detachable_50m_telemetry_keep 128 --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 --detachable_50m --detachable_50m_optimizer paged_adamw8bit --delta_every_steps 0 --delta_every_sec 0 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2787608 --batch_size 60 --dblock_blocks 14 --dblock_direct56_enabled 1 --dblock_layers_per_update 0 --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_full_anchor_every 128 --dblock_router_train_layer_policy cyclic --dblock_router_train_layers_per_update 1 --dblock_sublayer_mode off --dblock_target_route_activate_from_checkpoint 1 --dblock_target_route_activation_checkpoint_step 2717509 --dblock_target_route_activation_parent_pointer_sha256 04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827 --dblock_target_route_offset 1 --dblock_target_route_policy xor_fold56_v1 --dblock_train_layer_policy cyclic --dblock_train_layers_per_update 1 --dblock_train_sublayer_policy cyclic --dblock_train_sublayers_per_update 1 --granularity_label one-sublayer-local-grad-direct56-teacher-v27.16 --val_every_sec 600 --val_file /workspace/agillm43_heldout_v3/token_ids.json --val_sha256 513b30203ffd18167655deb2d9cd610541df87d1ea0b8a0f9241c58e68703d21",
473
+ "cwd": "/workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906",
474
+ "env": {
475
+ "AGILLM43_DBLOCK_CC_NOISE_SCALE": "0.0",
476
+ "AGILLM43_DBLOCK_CLEAN_CONTEXT": "e2e",
477
+ "AGILLM43_DBLOCK_DEPTH_ANCHOR_EVERY": "50",
478
+ "AGILLM43_DBLOCK_FULLSTACK_ANCHOR_CAP": "16376",
479
+ "AGILLM43_DBLOCK_FULLSTACK_AR_ROWS": "8",
480
+ "AGILLM43_DBLOCK_LOCAL_ROWS": "2",
481
+ "AGILLM43_DBLOCK_TARGET_ROUTE_ACTIVATE_FROM_CHECKPOINT": "1",
482
+ "AGILLM43_DBLOCK_TARGET_ROUTE_ACTIVATION_CHECKPOINT_STEP": "2717509",
483
+ "AGILLM43_DBLOCK_TARGET_ROUTE_ACTIVATION_PARENT_POINTER_SHA256": "04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827",
484
+ "AGILLM43_DBLOCK_TARGET_ROUTE_OFFSET": "1",
485
+ "AGILLM43_DBLOCK_TARGET_ROUTE_POLICY": "xor_fold56_v1",
486
+ "AGILLM43_DECOMPRESS_CACHE": "0",
487
+ "AGILLM43_DETACHABLE50M_ANCHOR_EVERY": "128",
488
+ "AGILLM43_DETACHABLE50M_ANCHOR_TOKENS": "<redacted>",
489
+ "AGILLM43_DETACHABLE50M_BATCH": "1",
490
+ "AGILLM43_DETACHABLE50M_CREATE_FROM_LEGACY": "1",
491
+ "AGILLM43_DETACHABLE50M_DISTILL_WEIGHT": "1.0",
492
+ "AGILLM43_DETACHABLE50M_ENABLED": "1",
493
+ "AGILLM43_DETACHABLE50M_EVERY": "1",
494
+ "AGILLM43_DETACHABLE50M_GATE_V2713_START_CHECKPOINT_SHA256": "39f442fc0e6f9fad6f875e067f63db67f33a41a53e8cbc49ce1591010bb4aec3",
495
+ "AGILLM43_DETACHABLE50M_HIDDEN_DISTILL_WEIGHT": "0.0",
496
+ "AGILLM43_DETACHABLE50M_LOGIT_DISTILL_ALPHA": "0.30",
497
+ "AGILLM43_DETACHABLE50M_LOGIT_DISTILL_TEMPERATURE": "1.0",
498
+ "AGILLM43_DETACHABLE50M_LOGIT_DISTILL_TOKENS": "<redacted>",
499
+ "AGILLM43_DETACHABLE50M_LOG_EVERY": "10",
500
+ "AGILLM43_DETACHABLE50M_LOSS_TOKENS": "<redacted>",
501
+ "AGILLM43_DETACHABLE50M_LR": "0.0001",
502
+ "AGILLM43_DETACHABLE50M_OPTIMIZER": "paged_adamw8bit",
503
+ "AGILLM43_DETACHABLE50M_SEED": "20260830",
504
+ "AGILLM43_DETACHABLE50M_SEED_FROM_MATURE": "1",
505
+ "AGILLM43_DETACHABLE50M_TEACHER_EVERY": "1",
506
+ "AGILLM43_DETACHABLE50M_TELEMETRY_KEEP": "128",
507
+ "AGILLM43_DETACHABLE50M_TOKENS": "<redacted>",
508
+ "AGILLM43_DETACHABLE50M_UPDATE_MODE": "sublayer",
509
+ "AGILLM43_DETACHABLE50M_UPGRADE_LOGIT_DISTILL": "1",
510
+ "AGILLM43_DETACHABLE50M_UPGRADE_MULTIMODE": "1",
511
+ "AGILLM43_DETACHABLE50M_UPGRADE_SAT_GATE_V2713": "1",
512
+ "AGILLM43_DETACHABLE50M_WEIGHT_DECAY": "0.0",
513
+ "AGILLM43_SEQUENCE_PREFETCH_DEPTH": "256",
514
+ "AGILLM43_SINGLEFILE_PROFILE": "intelligence-v25-satvar-pretrain",
515
+ "AGILLM43_SUPERVISE_SET_ARGS_JSON": "{\"--batch_size\": \"60\", \"--dblock_blocks\": \"14\", \"--dblock_direct56_enabled\": \"1\", \"--dblock_layers_per_update\": \"0\", \"--dblock_router\": \"shadow\", \"--dblock_router_blend\": \"0.0\", \"--dblock_router_full_anchor_every\": \"128\", \"--dblock_router_train_layer_policy\": \"cyclic\", \"--dblock_router_train_layers_per_update\": \"1\", \"--dblock_sublayer_mode\": \"off\", \"--dblock_target_route_activate_from_checkpoint\": \"1\", \"--dblock_target_route_activation_checkpoint_step\": \"2717509\", \"--dblock_target_route_activation_parent_pointer_sha256\": \"04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827\", \"--dblock_target_route_offset\": \"1\", \"--dblock_target_route_policy\": \"xor_fold56_v1\", \"--dblock_train_layer_policy\": \"cyclic\", \"--dblock_train_layers_per_update\": \"1\", \"--dblock_train_sublayer_policy\": \"cyclic\", \"--dblock_train_sublayers_per_update\": \"1\", \"--granularity_label\": \"one-sublayer-local-grad-direct56-teacher-v27.16\", \"--val_every_sec\": \"600\", \"--val_file\": \"/workspace/agillm43_heldout_v3/token_ids.json\", \"--val_sha256\": \"513b30203ffd18167655deb2d9cd610541df87d1ea0b8a0f9241c58e68703d21\"}",
516
+ "AGILLM43_WEIGHT_ACCEL_MODE": "shadow",
517
+ "AGILLM_DATASETS_STREAMING_READ_MAX_RETRIES": "2",
518
+ "AGILLM_DATASETS_STREAMING_READ_RETRY_INTERVAL": "1",
519
+ "AGILLM_DATASET_AGENT_ROUTER": "0",
520
+ "AGILLM_DATASET_HOTLOAD_CONFIG": "/workspace/agillm43_dataset_hotload.json",
521
+ "AGILLM_DATASET_HOT_RELOAD_CHECK_TOKENS": "<redacted>",
522
+ "AGILLM_DATASET_HOT_RELOAD_SEC": "5",
523
+ "AGILLM_DATASET_NN_ROUTER": "1",
524
+ "AGILLM_DATASET_ROUTER_BLEND": "0.25",
525
+ "AGILLM_DATASET_ROUTER_EXPLORE": "0.03",
526
+ "AGILLM_DATASET_ROUTER_LR": "0.005",
527
+ "AGILLM_DATASET_ROUTER_MIN_SCORE": "0.15",
528
+ "AGILLM_DATASET_ROUTER_SHARPNESS": "2.0",
529
+ "AGILLM_ENABLE_CUPY": "0",
530
+ "AGILLM_HF_HUB_DOWNLOAD_TIMEOUT": "30",
531
+ "AGILLM_HF_TRANSPORT": "geth_hf_relay",
532
+ "AGILLM_HOT_CONFIG": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
533
+ "AGILLM_HOT_CONFIG_SHA256": "",
534
+ "AGILLM_MAX_EXAMPLE_CHARS": "32768",
535
+ "AGILLM_MAX_EXAMPLE_TOKENS": "<redacted>",
536
+ "AGILLM_PARENT_POINTER_SHA256": "a944e39b2c75d1cdd48c9a575fdfaa158365edf6dbda546eba9ad7da8b4248cd",
537
+ "AGILLM_REMOTE_SHARDS_PER_ITER": "4",
538
+ "AGILLM_RUN_ID": "singlefile_intelligence_v25_20260906T181929Z_from_step000002787566",
539
+ "AGILLM_STREAM_NEXT_TIMEOUT_SEC": "60",
540
+ "AGILLM_STREAM_OPEN_TIMEOUT_SEC": "120",
541
+ "AGILLM_STREAM_SOCKET_TIMEOUT_SEC": "30",
542
+ "AGILLM_STREAM_SOURCE_FATAL_COOLDOWN_SEC": "600",
543
+ "AGILLM_STREAM_SOURCE_MAX_COOLDOWN_SEC": "120",
544
+ "AGILLM_SYNTHETIC_TOKENIZER": "<redacted>",
545
+ "CUDA_MODULE_LOADING": "LAZY",
546
+ "CUDA_VERSION": "12.1.1",
547
+ "CUDA_VISIBLE_DEVICES": "0",
548
+ "HF_HUB_DOWNLOAD_TIMEOUT": "30",
549
+ "HF_HUB_ETAG_TIMEOUT": "10",
550
+ "PYTORCH_CUDA_ALLOC_CONF": "backend:cudaMallocAsync",
551
+ "PYTORCH_VERSION": "2.2.1"
552
+ },
553
+ "parent_proc_cmdline": "/opt/conda/bin/python -u /workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906/agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.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",
554
+ "pid": 3798343,
555
+ "ppid": 3798273,
556
+ "proc_cmdline": "/opt/conda/bin/python -u /workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906/agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.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_schedule roundrobin --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --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_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 --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 --dblock_clean_context e2e --dblock_local_rows 2 --dblock_clean_context_noise_scale 0.0 --dblock_depth_anchor_every 50 --dblock_fullstack_anchor_cap 16376 --dblock_fullstack_ar_rows 8 --detachable_50m_upgrade_multimode --detachable_50m_upgrade_logit_distill --detachable_50m_upgrade_sat_gate_v2713 --detachable_50m_create_from_legacy --detachable_50m_seed_from_mature --detachable_50m_lr 0.0001 --detachable_50m_weight_decay 0.0 --detachable_50m_update_mode sublayer --detachable_50m_batch 1 --detachable_50m_tokens 64 --detachable_50m_loss_tokens 64 --detachable_50m_every 1 --detachable_50m_teacher_every 1 --detachable_50m_distill_weight 1.0 --detachable_50m_logit_distill_alpha 0.30 --detachable_50m_logit_distill_temperature 1.0 --detachable_50m_logit_distill_tokens 64 --detachable_50m_hidden_distill_weight 0.0 --detachable_50m_anchor_every 128 --detachable_50m_anchor_tokens 32 --detachable_50m_seed 20260830 --detachable_50m_log_every 10 --detachable_50m_telemetry_keep 128 --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 --detachable_50m --detachable_50m_optimizer paged_adamw8bit --delta_every_steps 0 --delta_every_sec 0 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2787608 --batch_size 60 --dblock_blocks 14 --dblock_direct56_enabled 1 --dblock_layers_per_update 0 --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_full_anchor_every 128 --dblock_router_train_layer_policy cyclic --dblock_router_train_layers_per_update 1 --dblock_sublayer_mode off --dblock_target_route_activate_from_checkpoint 1 --dblock_target_route_activation_checkpoint_step 2717509 --dblock_target_route_activation_parent_pointer_sha256 04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827 --dblock_target_route_offset 1 --dblock_target_route_policy xor_fold56_v1 --dblock_train_layer_policy cyclic --dblock_train_layers_per_update 1 --dblock_train_sublayer_policy cyclic --dblock_train_sublayers_per_update 1 --granularity_label one-sublayer-local-grad-direct56-teacher-v27.16 --val_every_sec 600 --val_file /workspace/agillm43_heldout_v3/token_ids.json --val_sha256 513b30203ffd18167655deb2d9cd610541df87d1ea0b8a0f9241c58e68703d21",
557
+ "schema": "agillm.launch.v1"
558
+ },
559
+ "local_step": 2801622,
560
+ "loss": 5.258305072784424,
561
+ "nat_mask_contract": "agillm.nat-mask.v1",
562
+ "nat_mask_contract_source": "explicit-legacy-recovery",
563
+ "nat_mask_eos_id": 1,
564
+ "nat_mask_legacy_blank_id": 1,
565
+ "nat_mask_migrated_from_id": 1,
566
+ "nat_mask_schema_version": 1,
567
+ "nat_mask_token_id": 2,
568
+ "pid": 3798343,
569
+ "schema_version": 1,
570
+ "seen_tok": 277974865920,
571
+ "source_is_lane_checkpoint": true,
572
+ "source_is_master_checkpoint": false,
573
+ "source_is_non_master_checkpoint": false,
574
+ "step": 2801622,
575
+ "train_argv": "/workspace/agillm43_production_v14/runtime_candidates/deepctx_v27_16_20260906/agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.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_schedule roundrobin --dblock_router_ramp_steps 64 --dblock_warmup_steps 14 --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_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 --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 --dblock_clean_context e2e --dblock_local_rows 2 --dblock_clean_context_noise_scale 0.0 --dblock_depth_anchor_every 50 --dblock_fullstack_anchor_cap 16376 --dblock_fullstack_ar_rows 8 --detachable_50m_upgrade_multimode --detachable_50m_upgrade_logit_distill --detachable_50m_upgrade_sat_gate_v2713 --detachable_50m_create_from_legacy --detachable_50m_seed_from_mature --detachable_50m_lr 0.0001 --detachable_50m_weight_decay 0.0 --detachable_50m_update_mode sublayer --detachable_50m_batch 1 --detachable_50m_tokens 64 --detachable_50m_loss_tokens 64 --detachable_50m_every 1 --detachable_50m_teacher_every 1 --detachable_50m_distill_weight 1.0 --detachable_50m_logit_distill_alpha 0.30 --detachable_50m_logit_distill_temperature 1.0 --detachable_50m_logit_distill_tokens 64 --detachable_50m_hidden_distill_weight 0.0 --detachable_50m_anchor_every 128 --detachable_50m_anchor_tokens 32 --detachable_50m_seed 20260830 --detachable_50m_log_every 10 --detachable_50m_telemetry_keep 128 --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 --detachable_50m --detachable_50m_optimizer paged_adamw8bit --delta_every_steps 0 --delta_every_sec 0 --resume /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt --save_dir /workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B --data_seed 2787608 --batch_size 60 --dblock_blocks 14 --dblock_direct56_enabled 1 --dblock_layers_per_update 0 --dblock_router shadow --dblock_router_blend 0.0 --dblock_router_full_anchor_every 128 --dblock_router_train_layer_policy cyclic --dblock_router_train_layers_per_update 1 --dblock_sublayer_mode off --dblock_target_route_activate_from_checkpoint 1 --dblock_target_route_activation_checkpoint_step 2717509 --dblock_target_route_activation_parent_pointer_sha256 04638cfbc19c0f9db6e6dfa03b2509d50a0bf3c56048a2cbd6e67154bc85c827 --dblock_target_route_offset 1 --dblock_target_route_policy xor_fold56_v1 --dblock_train_layer_policy cyclic --dblock_train_layers_per_update 1 --dblock_train_sublayer_policy cyclic --dblock_train_sublayers_per_update 1 --granularity_label one-sublayer-local-grad-direct56-teacher-v27.16 --val_every_sec 600 --val_file /workspace/agillm43_heldout_v3/token_ids.json --val_sha256 513b30203ffd18167655deb2d9cd610541df87d1ea0b8a0f9241c58e68703d21",
576
+ "train_script": "agillm43_singlefile_intelligence_v27_16_8_det50_packet_20260906.py",
577
+ "train_script_sha256": "2178bdeb5c6f18b89ba169a43ad9d5c0128e48c81d3111a7787ecc41462b9d8a",
578
+ "warmstart_base_seen_tok": 0,
579
+ "warmstart_base_step": 2787566,
580
+ "warmstart_kind": "warmstarted_from_lane_checkpoint",
581
+ "warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt",
582
+ "warmstart_source_step": 2787566
583
+ },
584
+ "alibi_mode": "corrected",
585
+ "alibi_scale": 0.0,
586
+ "checkpoint_summary": "warmstarted_from_lane_checkpoint; source=/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt; path=/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
587
+ "dataset_provenance": {
588
+ "chat": false,
589
+ "dataset_field_text": "text",
590
+ "dataset_hotload_config_path": "/workspace/agillm43_dataset_hotload.json",
591
+ "dataset_hotload_config_sha256": "61b8f59b84791d4dbee2a72a2ff4d0c9837f930f95cf94ab0013daf2abf1ae51",
592
+ "dataset_hotload_config_stamp": [
593
+ 50,
594
+ 1119951817,
595
+ 1788666957977200004,
596
+ 4227
597
+ ],
598
+ "dataset_hotload_config_used": true,
599
+ "dataset_hotload_enabled": true,
600
+ "dataset_hotload_generation": 10,
601
+ "dataset_hotload_schema": "agillm.dataset_hotload.config.v2",
602
+ "dataset_revision_map_sha256": "8361995a504a2365136b20e1a879ba78944a06309f13f3f013bbd0cdfc72225b",
603
+ "dataset_revisions": {
604
+ "EleutherAI/proof-pile-2": "901a9273a770e9d4138c5ddd91802f9c5c6cdc4b",
605
+ "HuggingFaceFW/fineweb": "9bb295ddab0e05d785b879661af7260fed5140fc",
606
+ "HuggingFaceFW/fineweb-edu": "87f09149ef4734204d70ed1d046ddc9ca3f2b8f9",
607
+ "HuggingFaceTB/cosmopedia": "0ae6ec63f91742bd2d1eaef4f02232c55d719385",
608
+ "HuggingFaceTB/finemath": "e92b25a616738fe95dc186b64dfb19f9c8525594",
609
+ "HuggingFaceTB/smollm-corpus": "3ba9d605774198c5868892d7a8deda78031a781f",
610
+ "Skylion007/openwebtext": "79d93d786212f7344586290adb811d4ae6a1762c",
611
+ "allenai/c4": "1588ec454efa1a09f29cd18ddd04fe05fc8653a2",
612
+ "allenai/dolma": "7f48140530a023e9ea4c5cfb141160922727d4d3",
613
+ "allenai/dolmino-mix-1124": "a319f19eef1e257417b11ea8c30da266ae175557",
614
+ "codeparrot/codeparrot-clean": "35a59fb025bc0a102f7d96eac09d145b896d487b",
615
+ "mlfoundations/dclm-baseline-1.0": "a3b142c183aebe5af344955ae20836eb34dcf69b",
616
+ "wikimedia/wikipedia": "b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
617
+ },
618
+ "dolma_payload_integrity": "external_url_manifest_pinned_payload_hashes_unavailable",
619
+ "dolma_url_list_sha256": "bc6ef505be920c94bb9e7147445638f606fdc3f5af622cd1d887dc48e21a2d5b",
620
+ "dolma_url_manifest_raw_sha256": "6fe8c40d51e5bebc9303453d7f533b232d69f7ae122123621a82621b1cd779b7",
621
+ "has_language_mix": true,
622
+ "has_numeracy": false,
623
+ "hot_config_expected_sha256": "",
624
+ "hot_config_has_append": true,
625
+ "hot_config_has_datasets": false,
626
+ "hot_config_mtime": 1788708887.095225,
627
+ "hot_config_path": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
628
+ "hot_config_used": true,
629
+ "hot_reload_capable": true,
630
+ "phase": "pretrain",
631
+ "proof_pile_lfs_manifest_sha256": "d020dfb796b02a75564e4e11d9e03185a55b3f9c6639f87830cfbe10737ed329",
632
+ "proof_pile_paths_sha256": "8a30018017e3b3d9a9bcfa06a3c404ccd113297b0b1d0c48e2e2aadf113e52b0",
633
+ "schema": "agillm.dataset_provenance.v1",
634
+ "sequence_prefetch_depth": 256,
635
+ "source_count": 16,
636
+ "source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96,wikimedia/wikipedia:20231101.en|1.63,allenai/c4:en|0.36,Skylion007/openwebtext|0.18,HuggingFaceTB/cosmopedia:web_samples_v2|0.64,HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65,EleutherAI/proof-pile-2:all|0.67,allenai/dolma:v1_6-sample|0.5,codeparrot/codeparrot-clean|0.24,HuggingFaceTB/finemath:finemath-4plus|0.32,HuggingFaceTB/finemath:infiwebmath-4plus|0.23,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05,allenai/dolmino-mix-1124|0.78,mlfoundations/dclm-baseline-1.0|1.05,HuggingFaceFW/fineweb-edu:sample-100BT|2.4,json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03",
637
+ "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",
638
+ "source_specs": [
639
+ "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96",
640
+ "wikimedia/wikipedia:20231101.en|1.63",
641
+ "allenai/c4:en|0.36",
642
+ "Skylion007/openwebtext|0.18",
643
+ "HuggingFaceTB/cosmopedia:web_samples_v2|0.64",
644
+ "HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65",
645
+ "EleutherAI/proof-pile-2:all|0.67",
646
+ "allenai/dolma:v1_6-sample|0.5",
647
+ "codeparrot/codeparrot-clean|0.24",
648
+ "HuggingFaceTB/finemath:finemath-4plus|0.32",
649
+ "HuggingFaceTB/finemath:infiwebmath-4plus|0.23",
650
+ "HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05",
651
+ "allenai/dolmino-mix-1124|0.78",
652
+ "mlfoundations/dclm-baseline-1.0|1.05",
653
+ "HuggingFaceFW/fineweb-edu:sample-100BT|2.4",
654
+ "json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03"
655
+ ],
656
+ "val_source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96,wikimedia/wikipedia:20231101.en|1.63,allenai/c4:en|0.36,Skylion007/openwebtext|0.18,HuggingFaceTB/cosmopedia:web_samples_v2|0.64,HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65,EleutherAI/proof-pile-2:all|0.67,allenai/dolma:v1_6-sample|0.5,codeparrot/codeparrot-clean|0.24,HuggingFaceTB/finemath:finemath-4plus|0.32,HuggingFaceTB/finemath:infiwebmath-4plus|0.23,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05,allenai/dolmino-mix-1124|0.78,mlfoundations/dclm-baseline-1.0|1.05,HuggingFaceFW/fineweb-edu:sample-100BT|2.4,json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03",
657
+ "val_source_requested": ""
658
+ },
659
+ "lr_schedule_origin_tok": 85002387456,
660
+ "nat_mask_contract": "agillm.nat-mask.v1",
661
+ "nat_mask_contract_source": "explicit-legacy-recovery",
662
+ "nat_mask_eos_id": 1,
663
+ "nat_mask_legacy_blank_id": 1,
664
+ "nat_mask_migrated_from_id": 1,
665
+ "nat_mask_schema_version": 1,
666
+ "nat_mask_token_id": 2,
667
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
668
+ "promotion_required": true,
669
+ "quality_gate_used_for_training": false,
670
+ "repair_schema": "",
671
+ "saved_at": "2026-09-08T05:31:16Z",
672
+ "schema": "agillm43.serving.candidate.v1",
673
+ "source_effective": "HuggingFaceFW/fineweb:CC-MAIN-2024-10|0.96,wikimedia/wikipedia:20231101.en|1.63,allenai/c4:en|0.36,Skylion007/openwebtext|0.18,HuggingFaceTB/cosmopedia:web_samples_v2|0.64,HuggingFaceTB/smollm-corpus:cosmopedia-v2|0.65,EleutherAI/proof-pile-2:all|0.67,allenai/dolma:v1_6-sample|0.5,codeparrot/codeparrot-clean|0.24,HuggingFaceTB/finemath:finemath-4plus|0.32,HuggingFaceTB/finemath:infiwebmath-4plus|0.23,HuggingFaceTB/smollm-corpus:fineweb-edu-dedup|1.05,allenai/dolmino-mix-1124|0.78,mlfoundations/dclm-baseline-1.0|1.05,HuggingFaceFW/fineweb-edu:sample-100BT|2.4,json:/workspace/agillm_wikitext103_raw_docs/train.jsonl|0.03",
674
+ "step": 2801622,
675
+ "training_pointer_schema": "agillm43.training.resume.v2",
676
+ "training_pointer_sha256": "57f1986e08084f2ee916bcb670e1f02320431d2d83ce56e4f9a1791d8017befb",
677
+ "warmstart_kind": "warmstarted_from_lane_checkpoint",
678
+ "warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02787566_deepctx_v27168_c0e8ba7500480fe2.pt"
679
+ }
checkpoints/step2801622_20260908/training_latest.json ADDED
@@ -0,0 +1,455 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "checkpoint_main_nbytes": 18131327,
3
+ "checkpoint_main_sha256": "a5b6079c3a056aada5e9a659d53c48414fa233f36f4a0a9f9a15fc19436db36a",
4
+ "checkpoint_name": "pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
5
+ "checkpoint_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
6
+ "continuation_state_schema": "agillm43.continuation.state.v2",
7
+ "continuation_state_sha256": "7ecd4d50d370a9d979236df09109fc3265f7baee21187f93f30905e794dc8189",
8
+ "dblock_committed_step": 1071271,
9
+ "dblock_target_route_clock": 84113,
10
+ "dblock_target_route_controller_source_sha256": "2178bdeb5c6f18b89ba169a43ad9d5c0128e48c81d3111a7787ecc41462b9d8a",
11
+ "dblock_target_route_last_committed_target": {
12
+ "attempts": 1,
13
+ "block": 2,
14
+ "canonical_index": 1,
15
+ "committed_step_after": 1071271,
16
+ "first_attempt_step": 84112,
17
+ "k": 1,
18
+ "last_attempt_step": 84112,
19
+ "layer_offset": 0,
20
+ "phase": 2,
21
+ "physical_layer": 4,
22
+ "q": 0,
23
+ "route_commit_clock": 84112,
24
+ "route_offset": 1,
25
+ "selection": {
26
+ "authority": "teacher",
27
+ "authority_granted": false,
28
+ "baseline_prediction": 5.883021058404964,
29
+ "candidate_hash": null,
30
+ "candidate_persisted": false,
31
+ "clock": 84112,
32
+ "fault_count": 0,
33
+ "guard_override": false,
34
+ "guard_reason": null,
35
+ "nominated": false,
36
+ "override": false,
37
+ "phase": 1,
38
+ "raw_prediction": 10,
39
+ "reason": "deterministic_teacher",
40
+ "schema": "agillm44.direct56.controller.v1",
41
+ "selected": 10,
42
+ "selected_target": 10,
43
+ "shadow_override": false,
44
+ "stage": "teacher_training",
45
+ "teacher_ce": 0.0003381395654287189,
46
+ "teacher_target": 10,
47
+ "teacher_training_receipts": 5264,
48
+ "teacher_value_prediction": 5.223366737365723,
49
+ "validation_samples": 0,
50
+ "value_prediction": 5.223366737365723
51
+ },
52
+ "sublayer": "ffn",
53
+ "target_id": 10,
54
+ "target_sha256": "6a3c66052f6a50f09e4e73606a51fc10a3c8207bfc6e5abbd871f296790884fa",
55
+ "teacher_target_id": 10
56
+ },
57
+ "dblock_target_route_offset": 1,
58
+ "dblock_target_route_payload_sha256": "0e98b884a2195c8b263e968583d3b01e408662b58a6c9012236e5183ac8fd724",
59
+ "dblock_target_route_pending_target": null,
60
+ "dblock_target_route_policy_id": "guarded-direct56-xor-fold56-v1",
61
+ "dblock_target_route_policy_sha256": "6dcf6e662a439ba6e4c275e43ec39452563813bf41ada7d683329a2a6f52c418",
62
+ "dblock_target_route_schema": "agillm43.dblock.target-route.direct56.v1",
63
+ "dblock_target_route_target_map_sha256": "31e64396588ebf72dd6aa20de6cbea14c20b6b0823269fade6f7471d37cc0836",
64
+ "dblock_target_route_topology_sha256": "c3b1679046220655820501bb30dc2733efc0628f661273ac3f2217355bdc2102",
65
+ "direct56_checkpointed": true,
66
+ "direct56_payload_digest": "76c735b5ddff670eed324062fc3b5ad606995fe28b65295068576c9304ed4ddc",
67
+ "direct56_schema": "agillm44.direct56.controller.v1",
68
+ "package_format": "block-sharded",
69
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt",
70
+ "quality_gate_used_for_training": false,
71
+ "record_sha256": "57f1986e08084f2ee916bcb670e1f02320431d2d83ce56e4f9a1791d8017befb",
72
+ "role": "training",
73
+ "saved_at": "2026-09-08T05:31:16Z",
74
+ "schema": "agillm43.training.resume.v2",
75
+ "seen_tok": 277974865920,
76
+ "shard_dir": "pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards",
77
+ "shards": [
78
+ {
79
+ "name": "core_base.pt",
80
+ "nbytes": 608544137,
81
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_base.pt",
82
+ "sha256": "85446a4d1e0f8a8da88f0ba587492b9392ecb9343b708b817d33dff458fae78e"
83
+ },
84
+ {
85
+ "name": "core_block_0000.pt",
86
+ "nbytes": 139913712,
87
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0000.pt",
88
+ "sha256": "a7c5c12e8f0c2c5187a46e57ea7364e005a8664433fc02eaa3e911ae25437d78"
89
+ },
90
+ {
91
+ "name": "core_block_0001.pt",
92
+ "nbytes": 140233029,
93
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0001.pt",
94
+ "sha256": "f29be97040e762565c9a1cf168b9d821c4ed9196939c852d4dd2bcd3de9ea250"
95
+ },
96
+ {
97
+ "name": "core_block_0002.pt",
98
+ "nbytes": 140221201,
99
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0002.pt",
100
+ "sha256": "2fc5ce8efcb34fb4ab161e352d8589008133ea464fd39471d4d8ed2565dccb12"
101
+ },
102
+ {
103
+ "name": "core_block_0003.pt",
104
+ "nbytes": 140206756,
105
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0003.pt",
106
+ "sha256": "6fe7158830cf4a8ff8d3f64780da3c363494983516bf1e69e594f9d65da1f1bd"
107
+ },
108
+ {
109
+ "name": "core_block_0004.pt",
110
+ "nbytes": 140361805,
111
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0004.pt",
112
+ "sha256": "da44c4b64d3e37ad0db59796853d85c6501d0edc16dbce45579e55f1712ba588"
113
+ },
114
+ {
115
+ "name": "core_block_0005.pt",
116
+ "nbytes": 140320473,
117
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0005.pt",
118
+ "sha256": "e8c9803ffa0e895d4a94362456e8c5df35b727ebd97fb4a5ac77bde54f7556ce"
119
+ },
120
+ {
121
+ "name": "core_block_0006.pt",
122
+ "nbytes": 140675228,
123
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0006.pt",
124
+ "sha256": "c11771e54ce92fc73d36a1534993000d54df68580d0098cfe53b115db3553a96"
125
+ },
126
+ {
127
+ "name": "core_block_0007.pt",
128
+ "nbytes": 140825651,
129
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0007.pt",
130
+ "sha256": "04963e9a3177ad21e6c4e26517bf8b274d6dfd6812c0d9858f2360ee596dba80"
131
+ },
132
+ {
133
+ "name": "core_block_0008.pt",
134
+ "nbytes": 140726377,
135
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0008.pt",
136
+ "sha256": "5b1e220a9b37659a78004c62312979064610b3d91c19fa2fcbf44be3481736bd"
137
+ },
138
+ {
139
+ "name": "core_block_0009.pt",
140
+ "nbytes": 140577080,
141
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0009.pt",
142
+ "sha256": "b759635a9dbb7866564144e391dd052301de48382d111d3aee2f36752f91750f"
143
+ },
144
+ {
145
+ "name": "core_block_0010.pt",
146
+ "nbytes": 140615020,
147
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0010.pt",
148
+ "sha256": "15316237e1c152a3706f6173a7ea1c5e52dd5da61fa505b3e2110e50f715cba4"
149
+ },
150
+ {
151
+ "name": "core_block_0011.pt",
152
+ "nbytes": 140713573,
153
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0011.pt",
154
+ "sha256": "05be4ec049c341f54532e61022980c782c7e5df660105602794b15d1fb2446a8"
155
+ },
156
+ {
157
+ "name": "core_block_0012.pt",
158
+ "nbytes": 140684568,
159
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0012.pt",
160
+ "sha256": "e15eae55f3c94fb14ba4f100c3f71d2c861dc227e76b119e353d93fc5f8fdb0e"
161
+ },
162
+ {
163
+ "name": "core_block_0013.pt",
164
+ "nbytes": 140494261,
165
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0013.pt",
166
+ "sha256": "5fc625af7965b03e84e628ddcd5264cdbb374e8922e7d098f347aa3014bc6b17"
167
+ },
168
+ {
169
+ "name": "core_block_0014.pt",
170
+ "nbytes": 140397330,
171
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0014.pt",
172
+ "sha256": "0abe5ecffb6707079d55e0d06f08aa441f6515c4d41bfceccb0fffe971683d3e"
173
+ },
174
+ {
175
+ "name": "core_block_0015.pt",
176
+ "nbytes": 140325682,
177
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0015.pt",
178
+ "sha256": "5b5f8522ae84f13a51c4b1fa6a33e4c4f20ec225b45e2d1b14c0c27bc18f9bd2"
179
+ },
180
+ {
181
+ "name": "core_block_0016.pt",
182
+ "nbytes": 140600921,
183
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0016.pt",
184
+ "sha256": "81d39c71b431ec4a00a72521324dca8ce4ad2385c23107f7e49246612d9440ba"
185
+ },
186
+ {
187
+ "name": "core_block_0017.pt",
188
+ "nbytes": 140265041,
189
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0017.pt",
190
+ "sha256": "b654b97a71d8d1715ed33c01688d456da14fa4e1eb6d0e7f42c92d97b6263c06"
191
+ },
192
+ {
193
+ "name": "core_block_0018.pt",
194
+ "nbytes": 140619172,
195
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0018.pt",
196
+ "sha256": "b024abd7a835d1168aa1084d1860f38289ffbd1ba80a70d6c2f77e44c4927694"
197
+ },
198
+ {
199
+ "name": "core_block_0019.pt",
200
+ "nbytes": 140402739,
201
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0019.pt",
202
+ "sha256": "636d5d9b19918231703a8dfd34e977d63ca330bf3c960fce12bc0d3334e841ab"
203
+ },
204
+ {
205
+ "name": "core_block_0020.pt",
206
+ "nbytes": 140491612,
207
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0020.pt",
208
+ "sha256": "76d7654ad869b8d2bb453439539032aa68c95db801f19a183ca664c254cafc91"
209
+ },
210
+ {
211
+ "name": "core_block_0021.pt",
212
+ "nbytes": 140272863,
213
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0021.pt",
214
+ "sha256": "9d7bed52092d573c26254503a74fafcf70a30c1cf7634e439ea134406555e4da"
215
+ },
216
+ {
217
+ "name": "core_block_0022.pt",
218
+ "nbytes": 140670831,
219
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0022.pt",
220
+ "sha256": "d107e7b216280457a582c369cb4cded00aeec05c254bb4f6f003f5789228b4b8"
221
+ },
222
+ {
223
+ "name": "core_block_0023.pt",
224
+ "nbytes": 140863573,
225
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0023.pt",
226
+ "sha256": "8a3bd18f86368ed15cc52aaf7b0e69cf35243706e75e4eb524f90d481cc8903f"
227
+ },
228
+ {
229
+ "name": "core_block_0024.pt",
230
+ "nbytes": 139702908,
231
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0024.pt",
232
+ "sha256": "4a6e03ffbf397f144b16e1937dd482050be9e9278db97f6ae9d37f70362269ce"
233
+ },
234
+ {
235
+ "name": "core_block_0025.pt",
236
+ "nbytes": 139721167,
237
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0025.pt",
238
+ "sha256": "30aa2b13ad1765a5b4d998d53c09ee87a9e21bf040ff0dde865361da4df76405"
239
+ },
240
+ {
241
+ "name": "core_block_0026.pt",
242
+ "nbytes": 139656163,
243
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0026.pt",
244
+ "sha256": "ad524db0606a45a0f23bc0214485a948c1b3d64cdc018337f558bfe74ef42d3f"
245
+ },
246
+ {
247
+ "name": "core_block_0027.pt",
248
+ "nbytes": 139836633,
249
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/core_block_0027.pt",
250
+ "sha256": "3af97e40d4d3493b681d3b4d145ae5ed79844c6f4f141d5a0ce3275731524059"
251
+ },
252
+ {
253
+ "name": "detachable_50m_base.pt",
254
+ "nbytes": 122156566,
255
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_base.pt",
256
+ "sha256": "0fd1b10f8be4019a4ded510ed690713965e90a9af394d0c6889039adfd1963d7"
257
+ },
258
+ {
259
+ "name": "detachable_50m_block_0000.pt",
260
+ "nbytes": 2721250,
261
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0000.pt",
262
+ "sha256": "24ed31cb62ea962a02882590ca86848106fcc21c8905d7a09fbc15b3c72d4f64"
263
+ },
264
+ {
265
+ "name": "detachable_50m_block_0001.pt",
266
+ "nbytes": 2699574,
267
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0001.pt",
268
+ "sha256": "63f88ae877a0be56a8bb3e8e397f0c99bb4b1f5e88e203bdc402c7d46d1080ad"
269
+ },
270
+ {
271
+ "name": "detachable_50m_block_0002.pt",
272
+ "nbytes": 2703373,
273
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0002.pt",
274
+ "sha256": "8ba0d67d5714f9bd0a1e0f6aae07babb4ea7c981586f901c7d7df8f1aed48044"
275
+ },
276
+ {
277
+ "name": "detachable_50m_block_0003.pt",
278
+ "nbytes": 2703473,
279
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0003.pt",
280
+ "sha256": "5e1124079b39bd5a71a9b127c3ab94d3a2fc3d12ec085c90aa8ce1b45149a685"
281
+ },
282
+ {
283
+ "name": "detachable_50m_block_0004.pt",
284
+ "nbytes": 2704887,
285
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0004.pt",
286
+ "sha256": "1a607fea423b75431b7c46eaf5a5b04d83e3345fab3194ffb58809bcdcea84b0"
287
+ },
288
+ {
289
+ "name": "detachable_50m_block_0005.pt",
290
+ "nbytes": 2710845,
291
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0005.pt",
292
+ "sha256": "200ea67dda87ef1733bc7b652eebf0a267726df0e54631f60f5ade9bacb83717"
293
+ },
294
+ {
295
+ "name": "detachable_50m_block_0006.pt",
296
+ "nbytes": 2720116,
297
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0006.pt",
298
+ "sha256": "76b20f4dc1495c183643c06d2d66fa8f9b35e62a8f8767cbfa8c7832788a3cf7"
299
+ },
300
+ {
301
+ "name": "detachable_50m_block_0007.pt",
302
+ "nbytes": 2708818,
303
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0007.pt",
304
+ "sha256": "09e379ea1d26b87b6d29d5addb7bc5e523de7c7b0e6a2e6f9b269f7a4785c669"
305
+ },
306
+ {
307
+ "name": "detachable_50m_block_0008.pt",
308
+ "nbytes": 2717718,
309
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0008.pt",
310
+ "sha256": "6ea75c27b7bfbb65468e7213b00a74a6861544e9e296bac9ebae9f68ed0cf433"
311
+ },
312
+ {
313
+ "name": "detachable_50m_block_0009.pt",
314
+ "nbytes": 2714125,
315
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0009.pt",
316
+ "sha256": "6918c6eef5e7e331ec2cf3946ffa1cf3fc653825b74b41f99584174121933b10"
317
+ },
318
+ {
319
+ "name": "detachable_50m_block_0010.pt",
320
+ "nbytes": 2708845,
321
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0010.pt",
322
+ "sha256": "61b75f7ef8d779b151419ec92b958811691d0b41ef54e05efb23efcc4e1521cd"
323
+ },
324
+ {
325
+ "name": "detachable_50m_block_0011.pt",
326
+ "nbytes": 2719712,
327
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0011.pt",
328
+ "sha256": "04c518a742a89ab5397585383845c19b2631caf56cd6296d7cc1f2b38f7e6830"
329
+ },
330
+ {
331
+ "name": "detachable_50m_block_0012.pt",
332
+ "nbytes": 2702588,
333
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0012.pt",
334
+ "sha256": "b5e721ad2686cbd6b2ee48e6a8a163a7245149ad15ee630da152c528f78da56d"
335
+ },
336
+ {
337
+ "name": "detachable_50m_block_0013.pt",
338
+ "nbytes": 2698065,
339
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0013.pt",
340
+ "sha256": "78cad066836dd69a2f1e8267985c58458f7412c26c59c4f6303b4cfbebd18d1d"
341
+ },
342
+ {
343
+ "name": "detachable_50m_block_0014.pt",
344
+ "nbytes": 2704295,
345
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0014.pt",
346
+ "sha256": "1ffea74f51445cab67e86ac503686e01dd69c630616e1a0f6d24b762a41a6226"
347
+ },
348
+ {
349
+ "name": "detachable_50m_block_0015.pt",
350
+ "nbytes": 2701042,
351
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0015.pt",
352
+ "sha256": "e23f8d8f3fcffe5177ad54b351da56165e5f806b2d12639c99cd50893f9cee66"
353
+ },
354
+ {
355
+ "name": "detachable_50m_block_0016.pt",
356
+ "nbytes": 2699076,
357
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0016.pt",
358
+ "sha256": "51cb2a48aa655f01030c86fb1d8379de2daf82db7c72fbfd890feb4660b782dd"
359
+ },
360
+ {
361
+ "name": "detachable_50m_block_0017.pt",
362
+ "nbytes": 2702513,
363
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0017.pt",
364
+ "sha256": "5d31637cc87a3c7f5a7a5e50545ce8e0f221e6c1ccf43f430eab4f5eba02de22"
365
+ },
366
+ {
367
+ "name": "detachable_50m_block_0018.pt",
368
+ "nbytes": 2708889,
369
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0018.pt",
370
+ "sha256": "46b704c9d9b269cf11a8688acb0477a5a14e2da7ad719ce4ba29f85824e41122"
371
+ },
372
+ {
373
+ "name": "detachable_50m_block_0019.pt",
374
+ "nbytes": 1970140,
375
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0019.pt",
376
+ "sha256": "d69953eb811204cd90ba5758daa8436c93b97d1098b156a92b8fad0850cf081d"
377
+ },
378
+ {
379
+ "name": "detachable_50m_block_0020.pt",
380
+ "nbytes": 1968775,
381
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0020.pt",
382
+ "sha256": "f19b2e245fc36ecddd43d33f579a52050a1fbab2dcf21e053b3f6302fca6562c"
383
+ },
384
+ {
385
+ "name": "detachable_50m_block_0021.pt",
386
+ "nbytes": 1969845,
387
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/detachable_50m_block_0021.pt",
388
+ "sha256": "cc73ce6a419793481988ad9bce7ffc6cf4b977baa8704c202a302ef5ea4e76d5"
389
+ },
390
+ {
391
+ "name": "ar.pt",
392
+ "nbytes": 608535164,
393
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/ar.pt",
394
+ "sha256": "f057d17f069b8c6862ef0c685eb27a7f76047a269b614a429c8a4fba51edc3d1"
395
+ },
396
+ {
397
+ "name": "sat.pt",
398
+ "nbytes": 608544477,
399
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/sat.pt",
400
+ "sha256": "682d64a2202854e6a7610c124f8414c38df675699a264d00494ed675b9f4d174"
401
+ },
402
+ {
403
+ "name": "nat.pt",
404
+ "nbytes": 608535164,
405
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/nat.pt",
406
+ "sha256": "f057d17f069b8c6862ef0c685eb27a7f76047a269b614a429c8a4fba51edc3d1"
407
+ },
408
+ {
409
+ "name": "opt.pt",
410
+ "nbytes": 1963167668,
411
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/opt.pt",
412
+ "sha256": "31ff6526abb49a615fb11c474579cefb5a4900f807626f5c17c2af5bd0b1cc85"
413
+ },
414
+ {
415
+ "name": "scaler.pt",
416
+ "nbytes": 149,
417
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/scaler.pt",
418
+ "sha256": "c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c"
419
+ },
420
+ {
421
+ "name": "agillm5_detachable_50m_bridge.pt",
422
+ "nbytes": 1224917,
423
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_bridge.pt",
424
+ "sha256": "b2b6861c163d9e0c0ad87cea7eaefa919bb2d54feac7f295fe5f691c34a50342"
425
+ },
426
+ {
427
+ "name": "agillm5_detachable_50m_opt.pt",
428
+ "nbytes": 66486594,
429
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_opt.pt",
430
+ "sha256": "10dac3ab92b29d9b8a801d17b5e972935bf05eea029d1383b491290f3d01d9ca"
431
+ },
432
+ {
433
+ "name": "agillm5_detachable_50m_scaler.pt",
434
+ "nbytes": 149,
435
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_scaler.pt",
436
+ "sha256": "c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c"
437
+ },
438
+ {
439
+ "name": "agillm5_detachable_50m_rng.pt",
440
+ "nbytes": 3541,
441
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_rng.pt",
442
+ "sha256": "450ef82bce6edd910ba868622d5bb6ef4ef9eeaa862cb85d0c78a605fa8359f5"
443
+ },
444
+ {
445
+ "name": "agillm5_detachable_50m_counters.pt",
446
+ "nbytes": 30526,
447
+ "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.shards/agillm5_detachable_50m_counters.pt",
448
+ "sha256": "dd7214f5a9cb2b9670b8340f0d89486b3fa8468d7c40d009ebfbdc014caa4c9c"
449
+ }
450
+ ],
451
+ "step": 2801622,
452
+ "tokenizer_sidecar_nbytes": 17465554,
453
+ "tokenizer_sidecar_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02801622_from02787566_20260908T053007_1788845407555765486_3798343_000108.pt.tokenizer.json",
454
+ "tokenizer_sidecar_sha256": "a50ab5a0ef074723d3fac3b7da79f915dcf06e49f8c9f4480feeabf837e03356"
455
+ }