checkpoint step2523907 (#321)

- 2d0bcf440b566c01bd8c08511321f3b48d07b126628f3da895b4d19e1d26929f (e7e677fc75514456d2cbcfa89aa9b6dea41dd8e2)
- bbd3f9a303f8d88824e9f30a98e7ebae7ffcd79457fbbd1142ab2fd9ca241321 (f3393e51902e8c059b7e2e0fe33f32a06eb8a6b1)
- 38a8a636fe7f8f4574f011988013d9b8928d27a37bd56efa38563f5ce42c35a6 (f7445ae2c5f3199146fa04dc8b8d16c0fef5c569)
- 00714680f16258dc8216367abc536d73eda42e98e3a3abeb20a4672b14257710 (102c1f5b117868e30a124f353f57ec3507993cef)
- 9fcb1508c970c1133b2dc0aa3e3c96c01c081a77b9426be4810f81f58fe4d318 (89c98f1591128f716d133f5e14c2038f4b24f229)

.gitattributes

@@ -354,3 +354,4 @@ checkpoints/step2520711_20260829/pretrain_step02520711_from02518657_20260829T212
 checkpoints/step2521509_20260829/pretrain_step02521509_from02518657_20260829T2153Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoints/step2522310_20260829/pretrain_step02522310_from02518657_20260829T2223Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoints/step2523112_20260829/pretrain_step02523112_from02518657_20260829T2253Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoints/step2523907_20260829/pretrain_step02523907_from02518657_20260829T2323Z.pt.tokenizer.json filter=lfs diff=lfs merge=lfs -text

checkpoints/step2523907_20260829/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c
+size 149

checkpoints/step2523907_20260829/serving_candidate.json

@@ -0,0 +1,546 @@
+{
+  "agillm43_provenance": {
+    "batch_size": 60,
+    "block_size": 2048,
+    "checkpoint_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt",
+    "checkpoint_type": "full",
+    "created_at_iso": "2026-08-29T23:23:09Z",
+    "created_at_unix": 1788045789.5822418,
+    "created_from_scratch": false,
+    "dataset_provenance": {
+      "chat": false,
+      "dataset_field_text": "text",
+      "dataset_hotload_config_path": "/workspace/agillm43_dataset_hotload.json",
+      "dataset_hotload_config_sha256": "2762ba666c280207bdb2293b6d382ef8077d167de79609e5bcc35d4499fd1527",
+      "dataset_hotload_config_stamp": [
+        73,
+        2196426857,
+        1787483694198261655,
+        1281
+      ],
+      "dataset_hotload_config_used": true,
+      "dataset_hotload_enabled": true,
+      "dataset_hotload_generation": 4,
+      "dataset_hotload_schema": "agillm.dataset_hotload.config.v2",
+      "dataset_revision_map_sha256": "8361995a504a2365136b20e1a879ba78944a06309f13f3f013bbd0cdfc72225b",
+      "dataset_revisions": {
+        "EleutherAI/proof-pile-2": "901a9273a770e9d4138c5ddd91802f9c5c6cdc4b",
+        "HuggingFaceFW/fineweb": "9bb295ddab0e05d785b879661af7260fed5140fc",
+        "HuggingFaceFW/fineweb-edu": "87f09149ef4734204d70ed1d046ddc9ca3f2b8f9",
+        "HuggingFaceTB/cosmopedia": "0ae6ec63f91742bd2d1eaef4f02232c55d719385",
+        "HuggingFaceTB/finemath": "e92b25a616738fe95dc186b64dfb19f9c8525594",
+        "HuggingFaceTB/smollm-corpus": "3ba9d605774198c5868892d7a8deda78031a781f",
+        "Skylion007/openwebtext": "79d93d786212f7344586290adb811d4ae6a1762c",
+        "allenai/c4": "1588ec454efa1a09f29cd18ddd04fe05fc8653a2",
+        "allenai/dolma": "7f48140530a023e9ea4c5cfb141160922727d4d3",
+        "allenai/dolmino-mix-1124": "a319f19eef1e257417b11ea8c30da266ae175557",
+        "codeparrot/codeparrot-clean": "35a59fb025bc0a102f7d96eac09d145b896d487b",
+        "mlfoundations/dclm-baseline-1.0": "a3b142c183aebe5af344955ae20836eb34dcf69b",
+        "wikimedia/wikipedia": "b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
+      },
+      "dolma_payload_integrity": "external_url_manifest_pinned_payload_hashes_unavailable",
+      "dolma_url_list_sha256": "bc6ef505be920c94bb9e7147445638f606fdc3f5af622cd1d887dc48e21a2d5b",
+      "dolma_url_manifest_raw_sha256": "6fe8c40d51e5bebc9303453d7f533b232d69f7ae122123621a82621b1cd779b7",
+      "has_language_mix": true,
+      "has_numeracy": false,
+      "hot_config_expected_sha256": "",
+      "hot_config_has_append": true,
+      "hot_config_has_datasets": false,
+      "hot_config_mtime": 1787722590.529335,
+      "hot_config_path": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
+      "hot_config_used": true,
+      "hot_reload_capable": true,
+      "phase": "pretrain",
+      "proof_pile_lfs_manifest_sha256": "d020dfb796b02a75564e4e11d9e03185a55b3f9c6639f87830cfbe10737ed329",
+      "proof_pile_paths_sha256": "8a30018017e3b3d9a9bcfa06a3c404ccd113297b0b1d0c48e2e2aadf113e52b0",
+      "schema": "agillm.dataset_provenance.v1",
+      "sequence_prefetch_depth": 256,
+      "source_count": 15,
+      "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",
+      "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",
+      "source_specs": [
+        "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"
+      ],
+      "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",
+      "val_source_requested": ""
+    },
+    "effective_global_step": 5042564,
+    "effective_seen_tok": 243853905920,
+    "global_origin_seen_tok": 0,
+    "global_origin_step": 2518657,
+    "gpu": {
+      "allocated_gb": 6.83,
+      "peak_allocated_gb": 18.89,
+      "reserved_gb": 22.53
+    },
+    "hostname": "ead6505f92c4",
+    "lane": "pretrain",
+    "launch": {
+      "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"
+      ],
+      "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",
+      "cwd": "/workspace/agillm43_production_v14/runtime",
+      "env": {
+        "AGILLM43_DECOMPRESS_CACHE": "0",
+        "AGILLM43_SEQUENCE_PREFETCH_DEPTH": "256",
+        "AGILLM43_SINGLEFILE_PROFILE": "intelligence-v25-satvar-pretrain",
+        "AGILLM_DATASETS_STREAMING_READ_MAX_RETRIES": "2",
+        "AGILLM_DATASETS_STREAMING_READ_RETRY_INTERVAL": "1",
+        "AGILLM_DATASET_AGENT_ROUTER": "0",
+        "AGILLM_DATASET_HOTLOAD_CONFIG": "/workspace/agillm43_dataset_hotload.json",
+        "AGILLM_DATASET_HOT_RELOAD_CHECK_TOKENS": "<redacted>",
+        "AGILLM_DATASET_HOT_RELOAD_SEC": "5",
+        "AGILLM_DATASET_NN_ROUTER": "1",
+        "AGILLM_DATASET_ROUTER_BLEND": "0.25",
+        "AGILLM_DATASET_ROUTER_EXPLORE": "0.03",
+        "AGILLM_DATASET_ROUTER_LR": "0.005",
+        "AGILLM_DATASET_ROUTER_MIN_SCORE": "0.15",
+        "AGILLM_DATASET_ROUTER_SHARPNESS": "2.0",
+        "AGILLM_ENABLE_CUPY": "0",
+        "AGILLM_HF_HUB_DOWNLOAD_TIMEOUT": "30",
+        "AGILLM_HF_TRANSPORT": "geth_hf_relay",
+        "AGILLM_HOT_CONFIG": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
+        "AGILLM_HOT_CONFIG_SHA256": "",
+        "AGILLM_MAX_EXAMPLE_CHARS": "32768",
+        "AGILLM_MAX_EXAMPLE_TOKENS": "<redacted>",
+        "AGILLM_PARENT_POINTER_SHA256": "0fb3c1b33756f72da78f5419606307366223602929633a581fb7431b6057dd69",
+        "AGILLM_REMOTE_SHARDS_PER_ITER": "4",
+        "AGILLM_RUN_ID": "singlefile_intelligence_v25_20260829T195836Z_from_step000002518657",
+        "AGILLM_STREAM_NEXT_TIMEOUT_SEC": "60",
+        "AGILLM_STREAM_OPEN_TIMEOUT_SEC": "120",
+        "AGILLM_STREAM_SOCKET_TIMEOUT_SEC": "30",
+        "AGILLM_STREAM_SOURCE_FATAL_COOLDOWN_SEC": "600",
+        "AGILLM_STREAM_SOURCE_MAX_COOLDOWN_SEC": "120",
+        "AGILLM_SYNTHETIC_TOKENIZER": "<redacted>",
+        "CUDA_MODULE_LOADING": "LAZY",
+        "CUDA_VERSION": "12.1.1",
+        "CUDA_VISIBLE_DEVICES": "0",
+        "HF_HUB_DOWNLOAD_TIMEOUT": "30",
+        "HF_HUB_ETAG_TIMEOUT": "10",
+        "PYTORCH_CUDA_ALLOC_CONF": "backend:cudaMallocAsync",
+        "PYTORCH_VERSION": "2.2.1"
+      },
+      "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",
+      "pid": 3402475,
+      "ppid": 3560703,
+      "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",
+      "schema": "agillm.launch.v1"
+    },
+    "local_step": 2523907,
+    "loss": 15.57977294921875,
+    "nat_mask_contract": "agillm.nat-mask.v1",
+    "nat_mask_contract_source": "explicit-legacy-recovery",
+    "nat_mask_eos_id": 1,
+    "nat_mask_legacy_blank_id": 1,
+    "nat_mask_migrated_from_id": 1,
+    "nat_mask_schema_version": 1,
+    "nat_mask_token_id": 2,
+    "pid": 3402475,
+    "schema_version": 1,
+    "seen_tok": 243853905920,
+    "source_is_lane_checkpoint": true,
+    "source_is_master_checkpoint": false,
+    "source_is_non_master_checkpoint": false,
+    "step": 2523907,
+    "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",
+    "train_script": "agillm43_singlefile_intelligence_v25_satvar_pretrain_final_20260826.py",
+    "train_script_sha256": "b7690fa26ac7e85f001068e93b0f170baec80d4f40f96241426c2a0582b7e00c",
+    "warmstart_base_seen_tok": 0,
+    "warmstart_base_step": 2518657,
+    "warmstart_kind": "warmstarted_from_lane_checkpoint",
+    "warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt",
+    "warmstart_source_step": 2518657
+  },
+  "alibi_mode": "corrected",
+  "alibi_scale": 0.0,
+  "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_step02523907_from02518657_20260829T2323Z.pt",
+  "dataset_provenance": {
+    "chat": false,
+    "dataset_field_text": "text",
+    "dataset_hotload_config_path": "/workspace/agillm43_dataset_hotload.json",
+    "dataset_hotload_config_sha256": "2762ba666c280207bdb2293b6d382ef8077d167de79609e5bcc35d4499fd1527",
+    "dataset_hotload_config_stamp": [
+      73,
+      2196426857,
+      1787483694198261655,
+      1281
+    ],
+    "dataset_hotload_config_used": true,
+    "dataset_hotload_enabled": true,
+    "dataset_hotload_generation": 4,
+    "dataset_hotload_schema": "agillm.dataset_hotload.config.v2",
+    "dataset_revision_map_sha256": "8361995a504a2365136b20e1a879ba78944a06309f13f3f013bbd0cdfc72225b",
+    "dataset_revisions": {
+      "EleutherAI/proof-pile-2": "901a9273a770e9d4138c5ddd91802f9c5c6cdc4b",
+      "HuggingFaceFW/fineweb": "9bb295ddab0e05d785b879661af7260fed5140fc",
+      "HuggingFaceFW/fineweb-edu": "87f09149ef4734204d70ed1d046ddc9ca3f2b8f9",
+      "HuggingFaceTB/cosmopedia": "0ae6ec63f91742bd2d1eaef4f02232c55d719385",
+      "HuggingFaceTB/finemath": "e92b25a616738fe95dc186b64dfb19f9c8525594",
+      "HuggingFaceTB/smollm-corpus": "3ba9d605774198c5868892d7a8deda78031a781f",
+      "Skylion007/openwebtext": "79d93d786212f7344586290adb811d4ae6a1762c",
+      "allenai/c4": "1588ec454efa1a09f29cd18ddd04fe05fc8653a2",
+      "allenai/dolma": "7f48140530a023e9ea4c5cfb141160922727d4d3",
+      "allenai/dolmino-mix-1124": "a319f19eef1e257417b11ea8c30da266ae175557",
+      "codeparrot/codeparrot-clean": "35a59fb025bc0a102f7d96eac09d145b896d487b",
+      "mlfoundations/dclm-baseline-1.0": "a3b142c183aebe5af344955ae20836eb34dcf69b",
+      "wikimedia/wikipedia": "b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
+    },
+    "dolma_payload_integrity": "external_url_manifest_pinned_payload_hashes_unavailable",
+    "dolma_url_list_sha256": "bc6ef505be920c94bb9e7147445638f606fdc3f5af622cd1d887dc48e21a2d5b",
+    "dolma_url_manifest_raw_sha256": "6fe8c40d51e5bebc9303453d7f533b232d69f7ae122123621a82621b1cd779b7",
+    "has_language_mix": true,
+    "has_numeracy": false,
+    "hot_config_expected_sha256": "",
+    "hot_config_has_append": true,
+    "hot_config_has_datasets": false,
+    "hot_config_mtime": 1787722590.529335,
+    "hot_config_path": "/workspace/agillm43_production_v14/contracts/hot_config_owner_v25.json",
+    "hot_config_used": true,
+    "hot_reload_capable": true,
+    "phase": "pretrain",
+    "proof_pile_lfs_manifest_sha256": "d020dfb796b02a75564e4e11d9e03185a55b3f9c6639f87830cfbe10737ed329",
+    "proof_pile_paths_sha256": "8a30018017e3b3d9a9bcfa06a3c404ccd113297b0b1d0c48e2e2aadf113e52b0",
+    "schema": "agillm.dataset_provenance.v1",
+    "sequence_prefetch_depth": 256,
+    "source_count": 15,
+    "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",
+    "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",
+    "source_specs": [
+      "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"
+    ],
+    "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",
+    "val_source_requested": ""
+  },
+  "lr_schedule_origin_tok": 85002387456,
+  "nat_mask_contract": "agillm.nat-mask.v1",
+  "nat_mask_contract_source": "explicit-legacy-recovery",
+  "nat_mask_eos_id": 1,
+  "nat_mask_legacy_blank_id": 1,
+  "nat_mask_migrated_from_id": 1,
+  "nat_mask_schema_version": 1,
+  "nat_mask_token_id": 2,
+  "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt",
+  "promotion_required": true,
+  "quality_gate_used_for_training": false,
+  "repair_schema": "",
+  "saved_at": "2026-08-29T23:23:58Z",
+  "schema": "agillm43.serving.candidate.v1",
+  "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",
+  "step": 2523907,
+  "training_pointer_schema": "agillm43.training.resume.v2",
+  "training_pointer_sha256": "b72c626f2da918d141e2c467ce5b27571a5e54f5388653ced969d4e3716c2215",
+  "warmstart_kind": "warmstarted_from_lane_checkpoint",
+  "warmstart_source_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02518657_from02477144_20260829T1953Z.pt"
+}

checkpoints/step2523907_20260829/training_latest.json

@@ -0,0 +1,228 @@
+{
+  "checkpoint_main_nbytes": 17669615,
+  "checkpoint_main_sha256": "44565cd157a54a6a40c8ee339153967ca5bbb4d090ce85bf0e35e79658c18185",
+  "checkpoint_name": "pretrain_step02523907_from02518657_20260829T2323Z.pt",
+  "checkpoint_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt",
+  "continuation_state_schema": "agillm43.continuation.state.v2",
+  "continuation_state_sha256": "a95afcb56cbbd43f32901928825099f25c141148d9dab5966e3dbba02181d724",
+  "dblock_committed_step": 793556,
+  "package_format": "block-sharded",
+  "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt",
+  "quality_gate_used_for_training": false,
+  "record_sha256": "b72c626f2da918d141e2c467ce5b27571a5e54f5388653ced969d4e3716c2215",
+  "role": "training",
+  "saved_at": "2026-08-29T23:23:58Z",
+  "schema": "agillm43.training.resume.v2",
+  "seen_tok": 243853905920,
+  "shard_dir": "pretrain_step02523907_from02518657_20260829T2323Z.pt.shards",
+  "shards": [
+    {
+      "name": "core_base.pt",
+      "nbytes": 608342765,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_base.pt",
+      "sha256": "95f6cd26fcb9e9272c122604bbf263e37c8b35301b60d5eec9bde4b363964da1"
+    },
+    {
+      "name": "core_block_0000.pt",
+      "nbytes": 117686532,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0000.pt",
+      "sha256": "e585b4769ba09a1e7f2e9f7d0127c7624e68d0f4c96744a44a447def8952237f"
+    },
+    {
+      "name": "core_block_0001.pt",
+      "nbytes": 116970176,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0001.pt",
+      "sha256": "b5c6ab688ab12469c7485acc1698ace4f792387397624f1dfc229bdc6412eaa7"
+    },
+    {
+      "name": "core_block_0002.pt",
+      "nbytes": 116948740,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0002.pt",
+      "sha256": "5a186edc1bbde628a9d272200f1732e3796504d0a6d492125decf30cfddbcdc7"
+    },
+    {
+      "name": "core_block_0003.pt",
+      "nbytes": 116924178,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0003.pt",
+      "sha256": "a0c6f5ba1e83caa423aac31683b4224577f4f1938064b4113a96f6d5e75196e9"
+    },
+    {
+      "name": "core_block_0004.pt",
+      "nbytes": 117018158,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0004.pt",
+      "sha256": "422a197f268b3a80078cd3059b0a54d84a066314252c5e588c56714879ca36c8"
+    },
+    {
+      "name": "core_block_0005.pt",
+      "nbytes": 116982525,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0005.pt",
+      "sha256": "d3b5fd597ef2a2c51c5ee327d7fa1fdcb1b18aa8f5e271812db7b917497a4f2f"
+    },
+    {
+      "name": "core_block_0006.pt",
+      "nbytes": 117366047,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0006.pt",
+      "sha256": "f7802d6730873eb1a4ecebfde476114a34bbf84ff3ea998ec831b7bddb03a120"
+    },
+    {
+      "name": "core_block_0007.pt",
+      "nbytes": 117604777,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0007.pt",
+      "sha256": "188ae9dae75f7c064643074ad7e048679832f7c4b486a569c56983e9f097930c"
+    },
+    {
+      "name": "core_block_0008.pt",
+      "nbytes": 117455285,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0008.pt",
+      "sha256": "dff08271b7315690ccf8427fe0cb72c8c15e94d33e0a9cd853444d2200616536"
+    },
+    {
+      "name": "core_block_0009.pt",
+      "nbytes": 117281897,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0009.pt",
+      "sha256": "34af75fff687eaa50e05ce8aea62cc93dd739c9c9277fe0eaec170d2a1f27771"
+    },
+    {
+      "name": "core_block_0010.pt",
+      "nbytes": 117411727,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0010.pt",
+      "sha256": "c431c8481b4e2e033a6e79c09aa0dbd8760c500089224bf16af968427da99461"
+    },
+    {
+      "name": "core_block_0011.pt",
+      "nbytes": 117420811,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0011.pt",
+      "sha256": "a488cd1bf8486ca09d25826191974c0300125f36f03b8863c100b5ada397eb7e"
+    },
+    {
+      "name": "core_block_0012.pt",
+      "nbytes": 123866147,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0012.pt",
+      "sha256": "6a3bfd223f90cb40a6b584fc1a1e33239d813f1edcffa7d039b6661be2658064"
+    },
+    {
+      "name": "core_block_0013.pt",
+      "nbytes": 120979568,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0013.pt",
+      "sha256": "05dff81430d124a2218b71a2fdeefb1824d7161c7589915011e97b5e4c4bff4d"
+    },
+    {
+      "name": "core_block_0014.pt",
+      "nbytes": 117790337,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0014.pt",
+      "sha256": "2629498d6414e0a3c3895c8bc7bed7e4021b77016a6c98c080a98793afae9814"
+    },
+    {
+      "name": "core_block_0015.pt",
+      "nbytes": 116987329,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0015.pt",
+      "sha256": "29a5ff89b772498def514b1203ba5847c63c211eefe69e2f768b53c86914eadb"
+    },
+    {
+      "name": "core_block_0016.pt",
+      "nbytes": 117234633,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0016.pt",
+      "sha256": "e73fe188505f8614117078e2c1d32c47dd382a9413e9439a70b3c34f52cd1094"
+    },
+    {
+      "name": "core_block_0017.pt",
+      "nbytes": 116966386,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0017.pt",
+      "sha256": "4f31469a2cfdd7cc3aafdde99303cd5f710841836f92b10d147b467987e314a9"
+    },
+    {
+      "name": "core_block_0018.pt",
+      "nbytes": 117109626,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0018.pt",
+      "sha256": "293e8f4713c1fde8dd57601c8d6f947d9ee920bad47ea860e3a95e89b2c7a706"
+    },
+    {
+      "name": "core_block_0019.pt",
+      "nbytes": 116995337,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0019.pt",
+      "sha256": "4e7171e90e35b0da8f84ebddc0e2af7edda9f6caa30b0aaf474b92847829df86"
+    },
+    {
+      "name": "core_block_0020.pt",
+      "nbytes": 117028830,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0020.pt",
+      "sha256": "2f9e5041bb9984a41e87f67a212840bc5432c65a5a155641d9642d624ba6bd32"
+    },
+    {
+      "name": "core_block_0021.pt",
+      "nbytes": 116871753,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0021.pt",
+      "sha256": "1478b0c2e98c5d09252f6922646938684d88edadff4781569a902739cde92b89"
+    },
+    {
+      "name": "core_block_0022.pt",
+      "nbytes": 117047254,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0022.pt",
+      "sha256": "3aedc94ab8be4f4e03f077fc8f664d108f4aa2e56f23f4fe98c803cb3dbc4f5e"
+    },
+    {
+      "name": "core_block_0023.pt",
+      "nbytes": 117316941,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0023.pt",
+      "sha256": "baeb30c44450147c74b7670e884be5935b62d76d422ccf8cab550b1914b20d61"
+    },
+    {
+      "name": "core_block_0024.pt",
+      "nbytes": 121464698,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0024.pt",
+      "sha256": "29da1577734c740d2a5fcdc5a24153bfb0652b71cdd4e0575bf35a354ed0690c"
+    },
+    {
+      "name": "core_block_0025.pt",
+      "nbytes": 121496241,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0025.pt",
+      "sha256": "2c40a5e35c60d748fa9c2f284666e0846394b6767913d9fc800a8c86c1010a7a"
+    },
+    {
+      "name": "core_block_0026.pt",
+      "nbytes": 121423912,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0026.pt",
+      "sha256": "8c52c524e15a13c21b26ad443001775d008792da38b0670be1ae7ff2da047295"
+    },
+    {
+      "name": "core_block_0027.pt",
+      "nbytes": 121616694,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/core_block_0027.pt",
+      "sha256": "aea553ce5aa10af124ac88c11b0a2aea8c03627ce5ee913c6240335d246f1a33"
+    },
+    {
+      "name": "ar.pt",
+      "nbytes": 608345522,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/ar.pt",
+      "sha256": "3e0782392e952f01b37c10b4ba7bb84e8bef32fffe2816b217ce002698d0de38"
+    },
+    {
+      "name": "sat.pt",
+      "nbytes": 608340680,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/sat.pt",
+      "sha256": "df5cc5db3fab573efe8e8ff32020e732ad76ea928a13abc2e8494047cee336e9"
+    },
+    {
+      "name": "nat.pt",
+      "nbytes": 608345512,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/nat.pt",
+      "sha256": "6ba279f8f51b18b6ac885da00ff0ad74fcb114596ee654b3e2661698036ccd1a"
+    },
+    {
+      "name": "opt.pt",
+      "nbytes": 1569676186,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/opt.pt",
+      "sha256": "4484c2c9778f85b502d04494a339a947c70bf68066c2371d3e20196f93e2b5e8"
+    },
+    {
+      "name": "scaler.pt",
+      "nbytes": 149,
+      "path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.shards/scaler.pt",
+      "sha256": "c2110dfd7b7a6c034fb62244a6f8f76a22d9936d55b6d0fe133e43e65b38631c"
+    }
+  ],
+  "step": 2523907,
+  "tokenizer_sidecar_nbytes": 17465482,
+  "tokenizer_sidecar_path": "/workspace/agillm43_production_v14/production_quality_v18_hard_spike_gates/checkpoints_continue400B_from200B/pretrain_step02523907_from02518657_20260829T2323Z.pt.tokenizer.json",
+  "tokenizer_sidecar_sha256": "c67cf8b65903569ad4f605aa98c474816dbbc28436a75188aa69d6e6681050ac"
+}