metadata
dataset_info:
features:
- name: conversations
list:
- name: content
dtype: string
- name: role
dtype: string
- name: instance_id
dtype: string
- name: traj_id
dtype: string
- name: improved
dtype: int64
- name: exec_feedback
struct:
- name: analysis
dtype: string
- name: metric_direction
dtype: string
- name: metric_value
dtype: float64
- name: status
dtype: string
- name: prev_best_metric
dtype: float64
- name: metric_diff
dtype: float64
splits:
- name: train
num_bytes: 47129301
num_examples: 204
download_size: 47126330
dataset_size: 47129301
configs:
- config_name: default
data_files:
- split: train
path: data/train-*