Spaces:
Running on Zero
Download README.md from hugging-apps/game2player: direct link, hf CLI and curl.
- Browser
- Download file 1.93 kB
-
https://huggingface.co/spaces/hugging-apps/game2player/resolve/main/README.md
- Command line
-
hf download hf://spaces/hugging-apps/game2player/README.md
-
curl -L -o README.md https://huggingface.co/spaces/hugging-apps/game2player/resolve/main/README.md
title: Game2Player
emoji: 🎮
colorFrom: red
colorTo: gray
sdk: gradio
sdk_version: 6.32.0
app_file: app.py
short_description: Visual game-playing agent — screenshot + task to next action
python_version: '3.12'
startup_duration_timeout: 30m
Game2Player — Visual Game-Playing Agent
Game2Player: Scaling Gameplay Experience for Visual Agents through Environment-Player Co-Development (paper · code)
This Space runs the released Game2Player-9B checkpoint (a Qwen3.5-9B fine-tune) with transformers, replicating the authors' serving recipe — bf16, ≤1,048,576 pixels per screenshot, thinking disabled, temperature 1.0 — and their PlayerBench prompt and action decoder, both ported 1:1 from the official repository.
Give the model a game screenshot and a task description (plus the game's rules
and allowed keys), and it returns the next keyboard or mouse action as a
computer_use tool call, decoded to a canonical action and drawn on the screenshot:
click coordinates, drag segments, key presses.
The bundled examples are real tasks from PlayerBench, the paper's benchmark:
screenshots captured from the benchmark's own games (107_simon-sequence and
105_memory-match, Apache-2.0), with the exact game-rule, role, and control strings
from the benchmark's task definitions.
Usage
- Pick an example, or upload your own screenshot and describe the task and rules.
- Hit Predict next action.
- The red crosshair on the output image is where the agent would click; the action summary line shows the decoded action (pixel and normalized 0–1000 coordinates).
The model is trained on a 1280×720 viewport; any input image is rescaled by the processor to the training pixel budget. Coordinates in the raw tool call are normalized 0–1000 and mapped back to the viewport by the decoder.