Joint training checkpoint step 10

README.md

@@ -10,7 +10,7 @@ tags: [tabular-data, masked-lm, token-classification, experimental]
 Experimental budget-limited mmBERT-base joint fine-tuning checkpoint.
 
 One shared encoder, 2 residual category BIO/insertion outputs (`text.encoding`, `text.spelling`), and the pretrained MLM vocabulary head. Declared schema rules detect deterministic violations before neural classification. Correction generates a complete cell inside `<replacement>` beside the visible `<original>` in the same cell. Valid copies are training targets too. The MLM head emits replacement tokens followed by `END_EDIT` and visible padding tokens.
-Completed optimizer updates: **0**. This checkpoint is a research baseline, not a validated automatic cleaning system.
+Completed optimizer updates: **10**. This checkpoint is a research baseline, not a validated automatic cleaning system.
 
 ## Use and resume
 
@@ -26,7 +26,7 @@ Dataset revision: `602b8a454ca448713d8a5658650d4feaad8d579c`. Train/validation f
 
 ## Validation results
 
-Selected step: 0. Stop reason: training in progress.
+Selected step: 10. Stop reason: training in progress.
 
 ```json
 {}
@@ -118,6 +118,16 @@ The learning rate warms up for 100 updates, then follows cosine decay to 10% of
     "best_score": 1.9120902847241386,
     "best_step": 0,
     "bad_checks": 0
-  }
+  },
+  "peak_gpu_bytes": 8804142080,
+  "step": 10,
+  "learning_rate": 2e-06,
+  "last_detection_loss": 1.4348130226135254,
+  "last_correction_loss": 1.137688398361206,
+  "training_elapsed_seconds": 29.446423814000013,
+  "exclusions": {},
+  "mean_detection_loss": 1.5767560005187988,
+  "mean_correction_loss": 6.01223161816597,
+  "loss_interval_updates": 8
 }
 ```

checkpoint.json

@@ -1,19 +1,19 @@
 {
   "version": 1,
-  "step": 0,
+  "step": 10,
   "dataset_revision": "602b8a454ca448713d8a5658650d4feaad8d579c",
   "base_revision": "c5955035435e2bf121cde7f3c8863ef52ff35d82",
   "files": {
-    "model.safetensors": "90b750926e48ae595b38bf48d9b890758a27de4ba5d1f0f00706bbf94b20e18d",
+    "model.safetensors": "847a5becfe2ab4cd93334554423287b8c4bfe043a000b9f2b3597d7a41acc27b",
     "config.json": "9e105ce4ed25ed2d6be2443eeb69932a530668a248ce3211d43a89a136d42531",
     "tokenizer_config.json": "51d83f5e4c1174f8dad62610076998730a57a9ac7025680653c720bca2a9b0a1",
     "tokenizer.json": "47e408472c05385fea5fb9f8b02bbd838d55da8f15a22c863a99fd31d89243f5",
-    "training_state.pt": "cd9e1448a8ac294f6f6d25dc2d5a1fc00faf3ff28d70475cf1942077b0ca5aae",
+    "training_state.pt": "19edc59bbd189bf5a012d140a8b04e1bcc31d39fb8386acecbe92c56e9b33a3c",
     "run_config.json": "d9e42266559b78f89751de2b1f9e9b27229cbd7646c126ea00fd691144495c1f",
-    "metrics.json": "10b5cd912f0fa835b3542364cc6ceb7a6a5232babdecfea8e1095dab06b6eab5",
+    "metrics.json": "f472d8706421824a3bb93cdb619ff787db6debc870301ad319ce74e252401143",
     "categories.json": "da00946a1a8dfd24dd3bd01f47a4493abe6ade281d6a8571ae9728e0406d4aab",
     "train_tabfix.py": "b16a43dbb276461d319a05bd4719de7520e0612fa23f7d83fb5fba58bebbc24f",
-    "README.md": "4b63d21bde39281a638ad6333175ebe4b8290646b9bf6434c0bff21c15673fcf",
+    "README.md": "2bbdaace0f2e661397797366a1767b292b23803883f668848588846477056d73",
     "LICENSE": "b6c1c7bb5aae03bb6b3396fc7c65cbe7306c422ed424b0b7b687d9a50cc15412",
     "BASE_MODEL_CARD.md": "35724415037ee3159aa733c16050620cc19a76219112a6ceeb3949572e0b37e9"
   },

metrics.json

@@ -73,5 +73,15 @@
     "best_score": 1.9120902847241386,
     "best_step": 0,
     "bad_checks": 0
-  }
+  },
+  "peak_gpu_bytes": 8804142080,
+  "step": 10,
+  "learning_rate": 2e-06,
+  "last_detection_loss": 1.4348130226135254,
+  "last_correction_loss": 1.137688398361206,
+  "training_elapsed_seconds": 29.446423814000013,
+  "exclusions": {},
+  "mean_detection_loss": 1.5767560005187988,
+  "mean_correction_loss": 6.01223161816597,
+  "loss_interval_updates": 8
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90b750926e48ae595b38bf48d9b890758a27de4ba5d1f0f00706bbf94b20e18d
+oid sha256:847a5becfe2ab4cd93334554423287b8c4bfe043a000b9f2b3597d7a41acc27b
 size 1231193844

training_state.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cd9e1448a8ac294f6f6d25dc2d5a1fc00faf3ff28d70475cf1942077b0ca5aae
-size 11293
+oid sha256:19edc59bbd189bf5a012d140a8b04e1bcc31d39fb8386acecbe92c56e9b33a3c
+size 2462488118