CoolFace
Datasetpublic

skkwowee/chimera-cs2

Chimera CS2 Dataset Labeled Counter-Strike 2 screenshots for vision-language model training. Each sample has: A CS2 gameplay screenshot Ground truth JSON with game_state, analysis, and advice Structure screenshots/ # PNG/JPG images labels/ # Matching JSON files (same stem name) manifest.jsonl # Data provenance tracking Usage from datasets import load_dataset ds = load_dataset("skkwowee/chimera-cs2") Stats Labels: 5309… See the full description on the dataset page: https://huggingface.co/datasets/skkwowee/chimera-cs2.

sourceHugging Facemitupdated 2mo agoView on Hugging Face
0likes3.8kdownloads
Dataset Card

Chimera CS2 Dataset

Labeled Counter-Strike 2 screenshots for vision-language model training.

Each sample has:

  • —A CS2 gameplay screenshot
  • —Ground truth JSON with game_state, analysis, and advice

Structure

screenshots/    # PNG/JPG images
labels/         # Matching JSON files (same stem name)
manifest.jsonl  # Data provenance tracking

Usage

python
from datasets import load_dataset
ds = load_dataset("skkwowee/chimera-cs2")

Stats

  • —Labels: 5309
  • —Screenshots: 4353

Paper

See, Then Think: Two-Phase VLM Training for Game Understanding

Code

github.com/skkwowee/chimera