Repository Public Research preview
-
Mon, Nov 9, 202613 commits
-
Single-term loss diagnostics presets
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Medium corpus v0 (13 372 docs / 371 419 eligible passages) summaries; tokenizer trainer on corpus passages; evidence summaries tracked
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
B2 T5-small baseline results (val, exploratory), predict_b2_hf script
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Denoising pretraining stage (unlabelled originals), --init-from fine-tuning, no-repeat n-gram decoding
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Medium corpus build script
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Protect BibTeX-key citations (protocol 1.1), standardised stylometry figure
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Paper figure/table/number generators with validated palette
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Eval suite runner, blind human-evaluation sheet generator
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Configurable bootstrap count, confirmatory flag, freeze script
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Gate B: pilot rewrites validated (5 383 pairs, 29.68 USD), versioned placeholder protocol, validator calibration, discipline taxonomy fix, STATUS, batch admin tooling
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
Add arXiv/LaTeX, Crossref/PDF and books adapters (agents), governance docs, KV-cache decoding, release/CLI/demo, B2 baseline, paper skeleton, pilot Gate B artefacts
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
-
WP-AI-0002: repository scaffold, licence governance, PMC S3 ingest, JATS parser, segmentation, dedup, splits, Claude rewrite engine (mock/live/batch), validation, HTR M1/M2/M3 models, trainer, evaluation, Gate A report, literature
···
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>