Datasets:
Tasks:
Text Generation
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
arxiv:2610.22312
benchmark
llm-evaluation
spatial-reasoning
state-prediction
executable-benchmark
License:
Download schema.json from xzx34/Grid-Bench: direct link, hf CLI and curl.
- Browser
- Download file 2.24 kB
-
https://huggingface.co/datasets/xzx34/Grid-Bench/resolve/main/schema.json
- Command line
-
hf download hf://datasets/xzx34/Grid-Bench/schema.json
-
curl -L -o schema.json https://huggingface.co/datasets/xzx34/Grid-Bench/resolve/main/schema.json
2.24 kB
| { | |
| "benchmark": { | |
| "cohort": "string", | |
| "dimension": "int64", | |
| "family": "string", | |
| "generator_version": "string", | |
| "gold_sha256": "string", | |
| "history_length": "int64", | |
| "in_main": "bool", | |
| "in_workload": "bool", | |
| "instance_id": "string", | |
| "instance_json": "string", | |
| "instance_sha256": "string", | |
| "prompt": "string", | |
| "prompt_sha256": "string", | |
| "prompt_version": "string", | |
| "public_id": "string", | |
| "public_trajectory_id": "string", | |
| "reference_json": "string", | |
| "rollout_id": "string", | |
| "schema_version": "string", | |
| "shape": "list_int64", | |
| "source_dataset": "string", | |
| "source_metadata_json": "string", | |
| "source_prompt_version": "string", | |
| "workload": "int64" | |
| }, | |
| "results": { | |
| "analysis_eligible": "bool", | |
| "answer_eligible": "bool", | |
| "answer_status": "string", | |
| "canonical_format": "bool", | |
| "cell_accuracy": "float64", | |
| "changed_cell_accuracy": "float64", | |
| "changed_cells": "int64", | |
| "cohort": "string", | |
| "collection_finalized": "bool", | |
| "copy_baseline_cell_accuracy": "float64", | |
| "copy_baseline_em": "bool", | |
| "em": "bool", | |
| "evaluation_protocol": "string", | |
| "extraction_applied": "bool", | |
| "extraction_method": "string", | |
| "finish_reason": "string", | |
| "history_length": "int64", | |
| "instance_id": "string", | |
| "latency_s": "float64", | |
| "model_id": "string", | |
| "model_name": "string", | |
| "model_order": "int64", | |
| "prompt_sha256": "string", | |
| "prompt_version": "string", | |
| "public_id": "string", | |
| "public_trajectory_id": "string", | |
| "refusal": "bool", | |
| "request_json": "string", | |
| "response_model": "string", | |
| "rollout_id": "string", | |
| "scoring_version": "string", | |
| "source_campaign": "string", | |
| "source_condition_id": "string", | |
| "source_dataset": "string", | |
| "source_request_sha256": "string", | |
| "strict_em": "bool", | |
| "termination": "string", | |
| "total_cells": "int64", | |
| "unchanged_cell_accuracy": "float64", | |
| "usage_complete": "bool", | |
| "usage_json": "string", | |
| "valid_json": "bool", | |
| "valid_state": "bool", | |
| "visible_text_empty": "bool", | |
| "wrong_cells": "int64", | |
| "zero_baseline_cell_accuracy": "float64", | |
| "zero_baseline_em": "bool" | |
| } | |
| } | |