README.md

---
dataset_info:
  features:
  - name: instance_id
    dtype: string
  - name: patch
    dtype: string
  - name: FAIL_TO_PASS
    list: string
  - name: PASS_TO_PASS
    list: string
  - name: image_name
    dtype: string
  - name: repo
    dtype: string
  - name: problem_statement
    dtype: string
  - name: task_id
    dtype: string
  - name: domain
    dtype: string
  - name: system_prompt
    dtype: string
  - name: user_prompt
    dtype: string
  - name: selected_tools
    list: string
  - name: restricted_tools
    list: 'null'
  - name: mcp_endpoint
    dtype: string
  - name: number_of_runs
    dtype: int64
  - name: reset_database_between_runs
    dtype: bool
  - name: gym_servers_config
    dtype: string
  - name: verifiers
    dtype: string
  - name: benchmark
    dtype: string
  - name: enterprise_ops_test_result
    dtype: string
  - name: skillsbench_test_result
    dtype: string
  - name: instruction
    dtype: string
  - name: num_actions
    dtype: int64
  - name: conversation_archive_path
    dtype: string
  - name: benchmark_run_name
    dtype: string
  - name: resolved
    dtype: bool
  - name: model_pred_patch
    dtype: string
  - name: model
    dtype: string
  - name: eval_report
    dtype: string
  - name: eval_test_output
    dtype: string
  - name: prompt_tokens
    dtype: int64
  - name: reasoning_tokens
    dtype: int64
  - name: completion_tokens
    dtype: int64
  splits:
  - name: valid
    num_bytes: 50306120
    num_examples: 379
  download_size: 21634283
  dataset_size: 50306120
configs:
- config_name: default
  data_files:
  - split: valid
    path: data/valid-*
---