erdem-erdem/maze_training_v1-9_tiny

---
dataset_info:
features:
- name: messages
list:
- name: content
dtype: string
- name: role
dtype: string
- name: maze_id
dtype: int64
- name: grid_size
dtype: int64
- name: step
dtype: int64
- name: total_steps
dtype: int64
splits:
- name: full_solution
num_bytes: 591672
num_examples: 1000
- name: multiturn_with_marks
num_bytes: 4363876
num_examples: 1000
- name: multiturn_past_moves
num_bytes: 4646592
num_examples: 1000
- name: multiturn_both
num_bytes: 4705592
num_examples: 1000
- name: multiturn_clean
num_bytes: 4304876
num_examples: 1000
- name: singleturn_with_marks
num_bytes: 9379084
num_examples: 14816
- name: singleturn_past_moves
num_bytes: 8847656
num_examples: 14816
- name: singleturn_both
num_bytes: 9721800
num_examples: 14816
- name: singleturn_clean
num_bytes: 8504940
num_examples: 14816
download_size: 5500268
dataset_size: 55066088
configs:
- config_name: default
data_files:
- split: full_solution
path: data/full_solution-*
- split: multiturn_with_marks
path: data/multiturn_with_marks-*
- split: multiturn_past_moves
path: data/multiturn_past_moves-*
- split: multiturn_both
path: data/multiturn_both-*
- split: multiturn_clean
path: data/multiturn_clean-*
- split: singleturn_with_marks
path: data/singleturn_with_marks-*
- split: singleturn_past_moves
path: data/singleturn_past_moves-*
- split: singleturn_both
path: data/singleturn_both-*
- split: singleturn_clean
path: data/singleturn_clean-*
---

This dataset contains maze-related dialogue data. Each record comprises a message list (including content and role fields), maze ID, grid size, current step count, and total step count. The dataset is categorized into multiple split types, including complete solutions, multi-turn dialogues (with markers, historical movement logs, a combination of the two, or a clean version), and single-turn dialogues (with markers, historical movement logs, a combination of the two, or a clean version). It is suitable for training and evaluating maze navigation and dialogue systems.