---
dataset_info:
- config_name: curves
  features:
  - name: run_id
    dtype: string
  - name: iteration
    dtype: int64
  - name: cumulative_tokens
    dtype: int64
  - name: best_val_score
    dtype: float64
  - name: accepted
    dtype: bool
  - name: candidate_idx
    dtype: int64
  - name: candidate_prompt
    dtype: string
  - name: best_prompt
    dtype: string
  splits:
  - name: train
    num_bytes: 72995
    num_examples: 6
  download_size: 26123
  dataset_size: 72995
- config_name: results
  features:
  - name: method
    dtype: string
  - name: k
    dtype: int64
  - name: max_iterations
    dtype: int64
  - name: max_metric_calls
    dtype: 'null'
  - name: best_val_score
    dtype: float64
  - name: num_candidates_explored
    dtype: int64
  - name: reflection_calls
    dtype: int64
  - name: reflection_input_tokens
    dtype: int64
  - name: reflection_output_tokens
    dtype: int64
  - name: reflection_total_tokens
    dtype: int64
  - name: reflection_cost_usd
    dtype: float64
  - name: wall_time_seconds
    dtype: float64
  - name: seed
    dtype: int64
  - name: task_lm
    dtype: string
  - name: reflection_lm
    dtype: string
  - name: panel
    dtype: string
  - name: test_accuracy
    dtype: float64
  - name: optimized_prompt
    dtype: string
  splits:
  - name: train
    num_bytes: 10422
    num_examples: 1
  download_size: 15367
  dataset_size: 10422
- config_name: traces
  features:
  - name: example_idx
    dtype: int64
  - name: input
    dtype: string
  - name: expected_answer
    dtype: string
  - name: model_response
    dtype: string
  - name: feedback
    dtype: string
  - name: score
    dtype: float64
  - name: correct
    dtype: bool
  - name: run_id
    dtype: string
  splits:
  - name: train
    num_bytes: 1503718
    num_examples: 150
  download_size: 657952
  dataset_size: 1503718
- config_name: val_traces
  features:
  - name: iteration
    dtype: int64
  - name: method
    dtype: string
  - name: k
    dtype: int64
  - name: candidate_idx
    dtype: int64
  - name: is_best_program
    dtype: bool
  - name: avg_val_score
    dtype: float64
  - name: example_idx
    dtype: int64
  - name: input
    dtype: string
  - name: expected_answer
    dtype: string
  - name: model_response
    dtype: string
  - name: score
    dtype: float64
  - name: correct
    dtype: bool
  - name: prompt_used
    dtype: string
  - name: cumulative_tokens
    dtype: int64
  - name: run_id
    dtype: string
  splits:
  - name: train
    num_bytes: 2464737
    num_examples: 180
  download_size: 683366
  dataset_size: 2464737
configs:
- config_name: curves
  data_files:
  - split: train
    path: curves/train-*
- config_name: results
  data_files:
  - split: train
    path: results/train-*
- config_name: traces
  data_files:
  - split: train
    path: traces/train-*
- config_name: val_traces
  data_files:
  - split: train
    path: val_traces/train-*
---