Update README.md

README.md

@@ -55,30 +55,9 @@ The GRPO format includes the tool prompt, terminal answer reward, source IDs,
 and training-only privileged hindsight. At inference/evaluation, expose only the
 problem and tool prompt to the policy; do not leak answer or hindsight fields.
 
-## Rename and reproducibility
-
-`rl_trace_opsd.parquet` is byte-for-byte identical to the historical
-`problem_difficulty_with_hindsight.parquet`:
-
-```text
-SHA256: 20a49e75f08e7ad0887839b69bcb84678038205682db8cd19c2b0f053c7589d7
-GRPO dataset digest: a07f1f45025e9bc909cca2db0b4c601c5cc53a4faf9d9ee6791341caee797f60
-```
-
-`grpo/manifest.json` retains its historical source filename and original bytes,
-because its hash is part of the dataset digest. `publication-manifest.json`
-records the rename separately. No split, row order, content, or ID was changed
-for publication. The pool retains 2,600 selected SFT-overlapping questions after
-the recorded overlap filtering; it is not completely disjoint from SFT data.
-
-For AutoDL, download the source into `data/numinamath/` and the `grpo/` directory
-into `data/`. Use `--difficulty-source data/numinamath/rl_trace_opsd.parquet` and
-`--dataset-dir data/grpo`.
-
 ## Source and attribution
 
 Derived from [NuminaMath-TIR](https://huggingface.co/datasets/AI-MO/NuminaMath-TIR),
 whose publisher declares Apache-2.0. ToolElite adds difficulty labels, solution
 hindsight, filtering, and tool-training formatting. Preserve upstream attribution
-and the included Apache license when redistributing. The provided solution and
-hindsight columns are released for training/research, not as benchmark inputs.
+and the included Apache license when redistributing.