|
Download data/processed/RQ1/GameBuilder-MCP/README.md from AINativeOps/AINativeBench: direct link, hf CLI and curl.
- Browser
- Download file 1.02 kB
-
https://huggingface.co/datasets/AINativeOps/AINativeBench/resolve/main/data/processed/RQ1/GameBuilder-MCP/README.md
- Command line
-
hf download hf://datasets/AINativeOps/AINativeBench/data/processed/RQ1/GameBuilder-MCP/README.md
-
curl -L -o README.md https://huggingface.co/datasets/AINativeOps/AINativeBench/resolve/main/data/processed/RQ1/GameBuilder-MCP/README.md
1.02 kB
GameBuilder-MCP (RQ1)
This folder contains analysis scripts and configuration files for evaluating the GameBuilder-MCP task results under the RESULTS/ directory.
What’s included
analyze_retry_patterns.py- Parses each session’s
execution_path.mdand aggregates retry/error statistics. - Outputs JSON/CSV summaries in this folder.
- Parses each session’s
evaluate_success.py- Computes per-model task success rates from
validation_result.json. - Outputs JSON/CSV summaries in this folder.
- Computes per-model task success rates from
evaluate_trajectory.py- Evaluates trajectory-level metrics by comparing parsed trajectories against a reference trajectory.
- Outputs CSV only.
reference_trajectory.yaml- Reference (ground-truth) trajectory definition and evaluation configuration.
Usage
Run scripts from this directory, for example:
python3 analyze_retry_patterns.py
python3 evaluate_success.py
python3 evaluate_trajectory.py --config reference_trajectory.yaml --format csv --output evaluation_results.csv