Publish verified OpenSysOne training snapshot pointer and model card

CURRENT_SNAPSHOT.json

@@ -1,14 +1,15 @@
 {
   "repo_id": "andyshu/opensysone",
-  "snapshot_id": "20260917T022835Z-hf-backup",
-  "path": "snapshots/20260917T022835Z-hf-backup",
-  "payload_commit": "e60eab96fdd568c5ce7c9e880640874bee63e88b",
-  "source_commit": "b156513bf6ac1475eaa24fd79e7beb94d159364f",
+  "snapshot_id": "20260917-expanded-pilot-hf-backup",
+  "path": "snapshots/20260917-expanded-pilot-hf-backup",
+  "payload_commit": "1a0a7fb679b88043c84325eadbbfc37c83345a31",
+  "source_commit": "fe7e48f806a7e947d689b9e966d38bc0dbedcc2e",
   "training_source_commits": [
+    "24b8ccf60d388f9cbb184e03a6ae260a1f5a8b86",
     "4a60423c39d70f8d50472ce4f4f7fa4a4bd9fce1"
   ],
-  "manifest_path": "publications/20260917T022835Z-hf-backup/b156513bf6ac1475eaa24fd79e7beb94d159364f/manifest.json",
-  "manifest_sha256": "5fc9dab067477b1d2e38c362bbdc3f45aa435cdcb800d3710d148f85cc217ce6",
-  "published_utc": "2026-09-17T03:59:39.663515+00:00",
+  "manifest_path": "publications/20260917-expanded-pilot-hf-backup/fe7e48f806a7e947d689b9e966d38bc0dbedcc2e/manifest.json",
+  "manifest_sha256": "9d6d8875cbd6c908db84a5cba2b7ac348df1c9d559c1f4836dedf198ec895fbc",
+  "published_utc": "2026-09-17T07:24:35.048421+00:00",
   "calibrated": false
 }

README.md

@@ -32,21 +32,25 @@ calibrated deployment have not completed. The original campaign ends on
 `FINAL_MODEL.json` will identify the separately verified calibrated release when
 finalization and upload succeed; its absence means no final release is recorded.
 
-Validation audit on 17 September, 02:10–02:15 UTC:
+Validation audit on 17 September, 07:00 UTC:
 
 | Selected candidate | Step | Validation accuracy | Crossfit selection NLL |
 | --- | ---: | ---: | ---: |
 | Qwen3 4B, main run | 2,500 | 93.55% | 0.188640 |
-| Qwen3 4B, lower learning rate | 2,500 | 92.58% | 0.218012 |
+| Qwen3 4B, lower learning rate | 4,000 | 93.75% | 0.190020 |
 | Qwen3.5 2B | 2,000 | 89.84% | 0.255294 |
+| Qwen3 4B, refinement | 1,500 | 94.73% | 0.170150 |
 
 All use the same 512 validation decisions. Selection uses four source-group-
 disjoint temperature-crossfit folds with fixed seed 431; smaller macro-family NLL
 is better. This is **validation-selection evidence**, not a held-out quality claim.
 The 2B run stopped cleanly at step 6,000 after eight evaluations without a new
-best. Both 4B runs continued, and another lower-rate 4B refinement was started
-from the main run's selected weights. See `source/NEXT_STEPS.md` for findings and
-why independent trials were retained over unmeasured distributed training.
+best. At 07:07 UTC the original GX10 4B run stopped cleanly at step 4,380,
+preserving its selected step 2,500 and full resumable state. GX10 now verifies an
+expanded-data candidate from the refinement's frozen selected step 1,500, while
+both Spark 4B runs continue. See `source/EXPANDED_DATA.md` for its latest run
+state, verification evidence and the broader training mix. Independent trials
+retain the original absolute deadlines.
 
 ## Contents and reconstruction
 
@@ -64,6 +68,8 @@ hosted inference application.
   and reconstruction metadata. These initial backups have no fitted temperature.
 - `snapshots/<id>/artifacts/<candidate>/checkpoint.pt`: resumable training state,
   including Adam and random states. Its step may be later than the selected best.
+- Expanded-data snapshots include transformed decision files and their source
+  notices, split hashes, diagnostic exclusions and original dataset lineage.
 - `final/<campaign>/`: final calibrated model and evaluation evidence, created only
   after successful finalization and publication.
 
@@ -91,7 +97,7 @@ Pinned bases, recorded as Apache-2.0 in their provenance:
 
 ## Data, evaluation and limitations
 
-Public training sources are SNLI, BoolQ, ARC and four-choice Banking77 routing.
+The original public training sources are SNLI, BoolQ, ARC and four-choice Banking77 routing.
 Social IQA is a wholly untrained task-family holdout. Frozen source pins, licences,
 raw/split hashes and group/deduplication audits are included in
 `source/results/public-decisions-v1-manifest.json`. That manifest credits
@@ -100,6 +106,17 @@ and records their original dataset licences. The original repository's
 `license: unknown` metadata is retained; no new licence for the project artifacts
 is assigned by this backup.
 
+The expanded candidate adds official training rows from HellaSwag (MIT), PIQA
+(AFL-3.0 according to its creator's pinned README) and CommonsenseQA (MIT).
+After the 512-token filter there are 80,765 training decisions, including all
+40,915 original retained examples. All four original reserved split files and
+their tokenized rows are unchanged. Another 383 retained new-source diagnostic
+decisions are excluded from both training and checkpoint selection. Source pins,
+credits, licences and transformations are in
+`source/results/20260917-expanded-data/dataset-manifest.json`. The unchanged
+selection set measures the original tasks; expansion alone is not evidence of
+better performance on the three added tasks.
+
 Validation selects checkpoints. Separate calibration fits one global temperature
 only after selection is frozen. Final test/holdout comparisons use the unchanged
 pretrained scorer, with a separately fitted base temperature and source-group
@@ -112,5 +129,7 @@ The harness implements local scoring, a loopback API, hosted Jev requests and
 response/timing comparisons. The local model identifies itself as OpenSysOne;
 compatibility with the request shape does not make it Jev. Local confidence is
 normalized entropy, not calibrated probability of correctness. Authenticated
-hosted Jev calls require `TYPESAFE_API_KEY` and have not been tested. Credentials,
-pretrained base weights and raw training datasets are not part of this backup.
+hosted Jev calls require `TYPESAFE_API_KEY` and have not been tested. Credentials
+and pretrained base weights are not part of this backup. Expanded-data snapshots
+may include the transformed training data, with upstream notices retained;
+raw upstream archives are referenced by pinned revision and checksum.