| --- |
| title: I See |
| emoji: 👀 |
| colorFrom: green |
| colorTo: blue |
| sdk: gradio |
| sdk_version: 4.44.1 |
| app_file: app.py |
| pinned: false |
| license: mit |
| --- |
| |
| # I See |
|
|
| An AI-involved multiple-choice I Spy game from ZygAI Research. |
|
|
| The app uses `meta-llama/Llama-3.1-8B-Instruct` as the game master. When the player starts, Llama generates all 10 clues, answer sets, correct answers, and explanations in one request. The game then plays from that cached set, which keeps the demo cost-effective. |
|
|
| ## Hugging Face Space Setup |
|
|
| 1. Create a Hugging Face Space with `Gradio` as the SDK. |
| 2. Upload `app.py`, `requirements.txt`, and this `README.md`. |
| 3. In the Space settings, add a secret named `HF_TOKEN`. |
| 4. Make sure that token has access to `meta-llama/Llama-3.1-8B-Instruct`. |
| 5. Restart the Space. |
|
|
| You can change the model by setting `HF_MODEL_ID` in the Space environment. |
|
|
| The app binds to Hugging Face's expected host and port automatically through the `PORT` environment variable. |
