Document completed engineering overfit check; production model remains pending

README.md

@@ -1,22 +1,25 @@
 ---
 license: apache-2.0
 base_model: convaiinnovations/laya-multilingual
-language:
-- en
-- zh
 tags:
-- clipboard
-- reranking
-- abstention
+- clipboard-ranking
+- cross-encoder
 - mlx
-- work-in-progress
+- research-in-progress
 ---
+
 # PasteWhat-Ranker-v1
 
-**Research in progress. No trained release or acceptance result is published yet.**
+**Research is in progress. No trained, calibrated and accepted production model is published here yet.**
+
+The project distills decision labels from `kimi-for-coding` into the original non-quantized Laya-multilingual encoder. New candidate and candidate-group-aware abstention heads score 1–20 existing clipboard entries. The model does not generate paste content or reasoning traces. Model inputs use application categories, not real app identities.
+
+The first local GPU engineering run is complete: 32 independently agent-reviewed Train examples were fitted in six epochs / 24 full-encoder updates, with all 32 training decisions correct. Conversion to MLX FP16 preserved those 32 decisions. This is **training-set fit, not generalization accuracy**, and that engineering checkpoint is not offered as a production model. Agent and teacher review is not human validation.
+
+Remaining production work includes the 5k pilot, 20k full training with three seeds, new-pool hard-example training, independent final MLX verification, calibration and frozen paired Test. Planned counts and acceptance targets are not reported as completed experiments. Train/Dev production, training, and Calibration/Test evaluation are owned by different agents with conceptual-family separation.
 
-Planned: decision-label distillation from `kimi-for-coding` into the non-quantized Laya-multilingual encoder, with candidate ranking and group-aware abstention heads. The model chooses existing clipboard content or returns null; it does not generate paste content.
+The final bundle will include PyTorch reference weights, MLX FP16 deployment weights, tokenizer, preprocessing, calibration policy, training/data manifests and measured quality/performance reports. Any unmet target will be disclosed.
 
-Training, independent synthetic evaluation, provenance and release status: [GitHub](https://github.com/mizorewww/pastewhat-ranker-v1).
+Source and ongoing execution records: [GitHub](https://github.com/mizorewww/pastewhat-ranker-v1). The [AppKit application](https://github.com/mizorewww/pastewhat) already supports Jev and has an adapter for the future calibrated local ranker. Synthetic-only metrics will not be claimed as real-user accuracy.
 
-Model artifacts will be published with tokenizer, preprocessing, calibration, measured metrics, and limitations. Synthetic-only results will not be presented as real-user accuracy.
+Apache-2.0; upstream source and conversion implementation attribution are documented in the source repository's LICENSE and NOTICE.