Joint training checkpoint step 2100

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: **1800**. This checkpoint is a research baseline, not a validated automatic cleaning system.
+Completed optimizer updates: **2100**. 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: 1800. Stop reason: training in progress.
+Selected step: 2100. Stop reason: training in progress.
 
 ```json
 {}
@@ -466,22 +466,81 @@ The learning rate warms up for 100 updates, then follows cosine decay to 10% of
         "3e8dff1e9e6db8d09c388b141194864e22c347550db958abeae4500015fa8ecb",
         "522ef97370ddbad950ea41bb8000b46488a7db7a05af8ee602117f2c0e2c400f"
       ]
+    },
+    {
+      "step": 2100,
+      "detection_loss": 0.012940295429871185,
+      "correction_loss": 0.49840453000904117,
+      "correction_full_mask_loss": 0.5877918274218246,
+      "correction_partial_mask_loss": 0.4090172325962577,
+      "selection_score": 0.04068216976788957,
+      "examples": 28,
+      "correction_edits": 52,
+      "correction_category_counts": {
+        "copy": 10,
+        "text.spelling": 7,
+        "format.boolean": 7,
+        "format.date": 3,
+        "text.encoding": 5,
+        "schema.category": 3,
+        "schema.identifier": 4,
+        "schema.type": 2,
+        "consistency.dependency": 6,
+        "consistency.temporal": 3,
+        "consistency.unit": 1,
+        "format.case": 1
+      },
+      "token_and_gap_tp_fp_fn_at_0.9": [
+        69,
+        0,
+        1
+      ],
+      "sample_ids": [
+        "7cc2b2069f314821114a2a2b09543c680923083980692c6a95b3371b316390d9-open-clean",
+        "2c30950b5459afe6967673e8b23c222adab49967bf80e72af0ef1d4fb96296ca-open",
+        "bd4bba00c41c73249e4fd9fe97876df42dbaefbe6aabeec4430d2acb63926318-open",
+        "0121e99febc488b49db6e39b07ef71255f5a94d9ef0292f7a34a705e59271bbe-open",
+        "4d74e65f621bbb1dfc2848cdb8aefbd32befa5042d8482c3340de2d9833c06ec-open",
+        "d120e4d54be4c9692441a06d9cc4bf1db02bd205c6a317a1a7ac4548c09666cc-open",
+        "d6b7a9c8fc0552cfcf0d26ce0ce1be2663d5aa86d80527fc77c4ccc02a036984-open",
+        "05f04c447b2d5e98f170add5fcdde123cb8db18946eb5e21ea8573fdfdd66ffe-open",
+        "b68fd4277af10beef7958e8d2f6b31805d635db888e60b38ae2306bf138ea584-open",
+        "c2710a557bc21c7f5d4c38d580e9d54042923e32d22848cc3f025caa864aa51c-open",
+        "da8d1a36641b16437fd72e9db23a60fbb6f185acdceec0f217491c232d3602b5-open-clean",
+        "55e92ca80f64845e208d3cda2f1e8fc3cb620ca57c66fce22d9bb9a86321af4a-open",
+        "53820fc3b14267390c15dfad0a80f83c9d0ed59c95e40c96c2283e4f3b167e09-open",
+        "d5f74ad40dca51caabef5cfcfc1845e3b40fcd71608304e4036786b6e27bfad6-open",
+        "6481cc1b3a6148ca5e8df1783aa3c15bd97b4224ea44ec304e2448bd613c3a07",
+        "8a1d338e2b1a352eac2c0834432588239c231b43df14e68402999151d62a70d8",
+        "04434b82c7685003521e12204c5af32606d402dd454be5d270c2b39c5a3c2b55",
+        "7aeaf4538d564e7bfc4a20cac6b97e42655170cd4ac40c98db0d24ff056ba388",
+        "9d7d44d25b2bd975829249dc3cd673c9a972f47ad8b1fa1bf0206f527d27934c",
+        "b01405d247c32e01b05e977fe963ca8ab440e1559f1c7f8adf0a99928d29b997",
+        "89f4e9132a30351d7836acfa80b56de26924f82d1b2403e48f1c265b4053ab99",
+        "fbe61f1cc2096a02443b68b5dbf6da5694e8cb60ec5fb445ca5bc29fc147b0e9",
+        "4dee644acbc9859035ebc82ce0d74ac45065b429f74ec833d9ae5dda3daf02a0",
+        "86e7d7c56f4e30ae4e63da4327d92c7a9ff2e18d47611dcba9e794b11a222f93",
+        "d515d8316e03100e37bb3be520d62c30463d7d30a235e0921c7ff2d88921c714",
+        "b4a926b6d1c343e53f4fd04dc327fb0fd86d3f4532b964fea22340ef73ca68de",
+        "3e8dff1e9e6db8d09c388b141194864e22c347550db958abeae4500015fa8ecb",
+        "522ef97370ddbad950ea41bb8000b46488a7db7a05af8ee602117f2c0e2c400f"
+      ]
     }
   ],
   "selection": {
     "best_score": 0.030781099261824835,
     "best_step": 1500,
-    "bad_checks": 1
+    "bad_checks": 2
   },
   "peak_gpu_bytes": 9905729024,
-  "step": 1800,
-  "learning_rate": 1.5167959675571553e-05,
-  "last_detection_loss": 9.324043276137672e-06,
-  "last_correction_loss": 0.05376574024558067,
-  "training_elapsed_seconds": 2975.924144399,
+  "step": 2100,
+  "learning_rate": 1.3566279527468294e-05,
+  "last_detection_loss": 1.30172784338356e-05,
+  "last_correction_loss": 0.04758565127849579,
+  "training_elapsed_seconds": 3456.865403115,
   "exclusions": {},
-  "mean_detection_loss": 1.2301624337851536e-05,
-  "mean_correction_loss": 0.0515645589432097,
+  "mean_detection_loss": 1.4484732309938408e-05,
+  "mean_correction_loss": 0.11099814534391043,
   "loss_interval_updates": 10
 }
 ```

checkpoint.json

@@ -1,19 +1,19 @@
 {
   "version": 1,
-  "step": 1800,
+  "step": 2100,
   "dataset_revision": "602b8a454ca448713d8a5658650d4feaad8d579c",
   "base_revision": "c5955035435e2bf121cde7f3c8863ef52ff35d82",
   "files": {
-    "model.safetensors": "4afcbe27ce59d540cbb4f78bc9fb6e3042fe3140a3a8973a5e750cfbd127d1d0",
+    "model.safetensors": "a7c419c2463a9c66e80591f5aa980d9d72a9eb7a1b29791e4c2ba4f73e4191c8",
     "config.json": "9e105ce4ed25ed2d6be2443eeb69932a530668a248ce3211d43a89a136d42531",
     "tokenizer_config.json": "51d83f5e4c1174f8dad62610076998730a57a9ac7025680653c720bca2a9b0a1",
     "tokenizer.json": "47e408472c05385fea5fb9f8b02bbd838d55da8f15a22c863a99fd31d89243f5",
-    "training_state.pt": "544f419fedc8bb58ad8a446ff57bb915577b3ef180bcdb907de2302fb31d4faa",
+    "training_state.pt": "2a64c903568f6f066db95df0b95f3f7c50f5b2785eb4a1ead03b0cf3a7ed322b",
     "run_config.json": "d9e42266559b78f89751de2b1f9e9b27229cbd7646c126ea00fd691144495c1f",
-    "metrics.json": "accf980241b2257a95e80dc365a4ffa0deb1fa65e815f1ae9c2af3f2d1d36e85",
+    "metrics.json": "33dc0436127c03541c7683593225c42574a3965c9a2eade81340d84cb5b930d9",
     "categories.json": "da00946a1a8dfd24dd3bd01f47a4493abe6ade281d6a8571ae9728e0406d4aab",
     "train_tabfix.py": "b16a43dbb276461d319a05bd4719de7520e0612fa23f7d83fb5fba58bebbc24f",
-    "README.md": "28b4f7b4db00231582391bb247c6ebdcf58892467a08dbae16d719594c211314",
+    "README.md": "6024911b2ab709655e2f2e9ed2081c2af5d53501ac58ad39d1d1dede82bb263c",
     "LICENSE": "b6c1c7bb5aae03bb6b3396fc7c65cbe7306c422ed424b0b7b687d9a50cc15412",
     "BASE_MODEL_CARD.md": "35724415037ee3159aa733c16050620cc19a76219112a6ceeb3949572e0b37e9"
   },

metrics.json

@@ -421,21 +421,80 @@
         "3e8dff1e9e6db8d09c388b141194864e22c347550db958abeae4500015fa8ecb",
         "522ef97370ddbad950ea41bb8000b46488a7db7a05af8ee602117f2c0e2c400f"
       ]
+    },
+    {
+      "step": 2100,
+      "detection_loss": 0.012940295429871185,
+      "correction_loss": 0.49840453000904117,
+      "correction_full_mask_loss": 0.5877918274218246,
+      "correction_partial_mask_loss": 0.4090172325962577,
+      "selection_score": 0.04068216976788957,
+      "examples": 28,
+      "correction_edits": 52,
+      "correction_category_counts": {
+        "copy": 10,
+        "text.spelling": 7,
+        "format.boolean": 7,
+        "format.date": 3,
+        "text.encoding": 5,
+        "schema.category": 3,
+        "schema.identifier": 4,
+        "schema.type": 2,
+        "consistency.dependency": 6,
+        "consistency.temporal": 3,
+        "consistency.unit": 1,
+        "format.case": 1
+      },
+      "token_and_gap_tp_fp_fn_at_0.9": [
+        69,
+        0,
+        1
+      ],
+      "sample_ids": [
+        "7cc2b2069f314821114a2a2b09543c680923083980692c6a95b3371b316390d9-open-clean",
+        "2c30950b5459afe6967673e8b23c222adab49967bf80e72af0ef1d4fb96296ca-open",
+        "bd4bba00c41c73249e4fd9fe97876df42dbaefbe6aabeec4430d2acb63926318-open",
+        "0121e99febc488b49db6e39b07ef71255f5a94d9ef0292f7a34a705e59271bbe-open",
+        "4d74e65f621bbb1dfc2848cdb8aefbd32befa5042d8482c3340de2d9833c06ec-open",
+        "d120e4d54be4c9692441a06d9cc4bf1db02bd205c6a317a1a7ac4548c09666cc-open",
+        "d6b7a9c8fc0552cfcf0d26ce0ce1be2663d5aa86d80527fc77c4ccc02a036984-open",
+        "05f04c447b2d5e98f170add5fcdde123cb8db18946eb5e21ea8573fdfdd66ffe-open",
+        "b68fd4277af10beef7958e8d2f6b31805d635db888e60b38ae2306bf138ea584-open",
+        "c2710a557bc21c7f5d4c38d580e9d54042923e32d22848cc3f025caa864aa51c-open",
+        "da8d1a36641b16437fd72e9db23a60fbb6f185acdceec0f217491c232d3602b5-open-clean",
+        "55e92ca80f64845e208d3cda2f1e8fc3cb620ca57c66fce22d9bb9a86321af4a-open",
+        "53820fc3b14267390c15dfad0a80f83c9d0ed59c95e40c96c2283e4f3b167e09-open",
+        "d5f74ad40dca51caabef5cfcfc1845e3b40fcd71608304e4036786b6e27bfad6-open",
+        "6481cc1b3a6148ca5e8df1783aa3c15bd97b4224ea44ec304e2448bd613c3a07",
+        "8a1d338e2b1a352eac2c0834432588239c231b43df14e68402999151d62a70d8",
+        "04434b82c7685003521e12204c5af32606d402dd454be5d270c2b39c5a3c2b55",
+        "7aeaf4538d564e7bfc4a20cac6b97e42655170cd4ac40c98db0d24ff056ba388",
+        "9d7d44d25b2bd975829249dc3cd673c9a972f47ad8b1fa1bf0206f527d27934c",
+        "b01405d247c32e01b05e977fe963ca8ab440e1559f1c7f8adf0a99928d29b997",
+        "89f4e9132a30351d7836acfa80b56de26924f82d1b2403e48f1c265b4053ab99",
+        "fbe61f1cc2096a02443b68b5dbf6da5694e8cb60ec5fb445ca5bc29fc147b0e9",
+        "4dee644acbc9859035ebc82ce0d74ac45065b429f74ec833d9ae5dda3daf02a0",
+        "86e7d7c56f4e30ae4e63da4327d92c7a9ff2e18d47611dcba9e794b11a222f93",
+        "d515d8316e03100e37bb3be520d62c30463d7d30a235e0921c7ff2d88921c714",
+        "b4a926b6d1c343e53f4fd04dc327fb0fd86d3f4532b964fea22340ef73ca68de",
+        "3e8dff1e9e6db8d09c388b141194864e22c347550db958abeae4500015fa8ecb",
+        "522ef97370ddbad950ea41bb8000b46488a7db7a05af8ee602117f2c0e2c400f"
+      ]
     }
   ],
   "selection": {
     "best_score": 0.030781099261824835,
     "best_step": 1500,
-    "bad_checks": 1
+    "bad_checks": 2
   },
   "peak_gpu_bytes": 9905729024,
-  "step": 1800,
-  "learning_rate": 1.5167959675571553e-05,
-  "last_detection_loss": 9.324043276137672e-06,
-  "last_correction_loss": 0.05376574024558067,
-  "training_elapsed_seconds": 2975.924144399,
+  "step": 2100,
+  "learning_rate": 1.3566279527468294e-05,
+  "last_detection_loss": 1.30172784338356e-05,
+  "last_correction_loss": 0.04758565127849579,
+  "training_elapsed_seconds": 3456.865403115,
   "exclusions": {},
-  "mean_detection_loss": 1.2301624337851536e-05,
-  "mean_correction_loss": 0.0515645589432097,
+  "mean_detection_loss": 1.4484732309938408e-05,
+  "mean_correction_loss": 0.11099814534391043,
   "loss_interval_updates": 10
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4afcbe27ce59d540cbb4f78bc9fb6e3042fe3140a3a8973a5e750cfbd127d1d0
+oid sha256:a7c419c2463a9c66e80591f5aa980d9d72a9eb7a1b29791e4c2ba4f73e4191c8
 size 1231193844

training_state.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:544f419fedc8bb58ad8a446ff57bb915577b3ef180bcdb907de2302fb31d4faa
+oid sha256:2a64c903568f6f066db95df0b95f3f7c50f5b2785eb4a1ead03b0cf3a7ed322b
 size 2462488118