README.md

metadata

pretty_name: TCS Direction MC Value - Solution Label
license: apache-2.0
language:
  - en
task_categories:
  - text-classification
tags:
  - process-reward-model
  - value-model
  - mathematical-reasoning
  - theorem-proving
  - monte-carlo
  - tree-search
  - ablation
configs:
  - config_name: default
    data_files:
      - split: train
        path: train.parquet
      - split: validation
        path: validation.parquet

TCS Direction MC Value: Solution Label

Partial rows include 3 same-problem endpoint solution references; reference reward labels are included. Endpoint rows use the explicit no-context sentinel.

Each row is a two-message conversation ending in the literal assistant target yes. Training uses the dense reward column as the soft target P(correct); correct is only the legacy boolean projection. Hidden model reasoning is not included in endpoint state.

Identity

• Source: asingh15/tcs-qwen36-27b-direction-rollouts-partial-1017-mc8
• Source revision: 685b614552d9ba5e2f27d6aeee1ff9d6e83c9635
• Run ID: tcs_qwen36_27b_direction_beam_full1801_mc8_20260815
• Arm: solution_label
• Verification file SHA-256: 7ddab749f98b251aea5d3ebcdb4137fd579a6858854f756f60ea4043dd032e38
• Verification contract SHA-256: 329c0fc6e175e3baf139214bee239fe956ef4158a598ee50d21bbcdb21c77279
• Maximum observed tokens: 29650
• Validated token cap: 49152

Splits

Split	Rows	Bytes	SHA-256
train	32,004	65,906,540	8f3d221bdebbbb861990705b6afaf855912a087d7e39ebae74ba34f59a6660a0
validation	4,608	9,403,527	ad28e83723245a3ead1f4cb87aeb4a86efec4c689f39effaea8324b170f690ee

dataset_provenance.json, verification.json, and _SUCCESS.json pin the builder contract, source artifacts, cross-arm row alignment, and published file identities.