Ghost Hunter RLHF: 8-Bit FPS Gameplay Screenshots for Preference Modeling

Ghosthunter RLHF contains under 1,000 gameplay screenshots from the 8-bit first-person shooter "Ghost Hunter," developed by webxos and updated in March 2026. The collection captures specific instances of successful ghost elimination using precision auto-fire to facilitate reinforcement learning from human feedback (RLHF).

Use Cases

• Training a preference model to identify 'good' shots based on the precision auto-fire screenshots
• Developing visual classification models to detect 8-bit ghost entities within game frames
• Fine-tuning vision-language models for game state description using the multimodal gameplay data

Strengths

• Includes a functional OpenAI Gym environment in the /gym/ folder for custom data generation
• MIT licensed for open redistribution and modification
• Highly specific focus on precision auto-fire mechanics in an 8-bit environment

Limitations

• Small sample size of fewer than 1,000 records
• Limited to a single 8-bit game environment with low visual complexity
• Lack of explicit negative preference labels in the base image set

Provenance

Source

webxos via Hugging Face

Collection Method

Captured during active gameplay of the Ghost Hunter 8-bit FPS

Freshness

Last updated March 2026.