PredictLM v11.0 + Mini ship-bundle

README.md

@@ -107,7 +107,7 @@ PredictLM's TTT is an independent implementation of the published technique. Thi
 
 - **Developed by**: ZeroOne Research
 - **Distilled from**: [predictlm-base-26m](https://huggingface.co/zerooneresearch/predictlm-base-26m) (v11.0)
-- **Model card contact**: open an issue at the [code repo](https://github.com/zerooneresearch/predictlm-v11) or message the org on the Hub
+- **Model card contact**: message the org on the Hub
 - **License**: Apache 2.0 — permissive, commercial use allowed
 
 ## Why Mini (when to prefer this over Base)
@@ -201,8 +201,6 @@ Mini was trained via **warm-start sliced distillation**: a novel recipe for comp
 
 The critical insight: distillation from scratch (Option A in our experiments) **failed to transfer to real OpenML data** — student matched teacher on synthetic but couldn't generalize. Warm-start sliced distillation (Option B, this release) succeeded because the student inherits the teacher's transfer ability as the starting point; distillation only needs to refine.
 
-Reproducible from scratch with `scripts/train_v11_06_tiny.py` in the code repo.
-
 ## Intended use, limitations, ethical considerations
 
 Identical to [predictlm-base-26m](https://huggingface.co/zerooneresearch/predictlm-base-26m) — see that model card for full details:
@@ -222,30 +220,8 @@ The known weaknesses (cls below XGBoost; below TabPFN-2.5 / TabICLv2 on both axe
 - **Training step**: 30,000 (final)
 - **Training seed**: 42
 - **Teacher**: `predictlm-base-26m` (v11.0)
-- **Distillation recipe**: warm-start slice + online KL distillation, see `scripts/train_v11_06_tiny.py`
+- **Distillation recipe**: warm-start slice + online KL distillation
 - **Eval-lock manifest SHA-256**: `fe4da8cccfc78fc3c7746579f604154af7d37e525c4fd575965ba77ce4fe0841` (identical to Base)
-- **Code**: pinned at the v11.0 release tag of [zerooneresearch/predictlm-v11](https://github.com/zerooneresearch/predictlm-v11)
-
-To reproduce from scratch:
-
-```bash
-# Pull the v11.0 teacher
-huggingface-cli download zerooneresearch/predictlm-base-26m v11_final.pt --local-dir ./
-
-# Reproduce Mini
-python3 scripts/train_v11_06_tiny.py \
-    --teacher-ckpt v11_final.pt \
-    --warm-start-from-v11 v11_final.pt \
-    --mlp-variant gelu --norm-variant layernorm --share-factor 2 \
-    --corpus-manifest data/v11_05_corpus_manifest.json \
-    --copula-dir data/copulas \
-    --steps 30000 --batch-size 1 --accum-steps 8 \
-    --lr 3e-5 --min-lr 3e-6 --warmup-steps 500 \
-    --n-context-max 256 --n-query-max 64 \
-    --device cuda --precision fp16 \
-    --probe-interval 2500 --probe-warmup-steps 0 --probe-floor 0.30 \
-    --out-dir runs/v11_06_tiny_reproduce --seed 42
-```
 
 ## Licensing