Spaces:
Sleeping
Sleeping
| title: Token Probability Visualization API | |
| emoji: 📊 | |
| colorFrom: blue | |
| colorTo: purple | |
| sdk: gradio | |
| sdk_version: 4.44.0 | |
| app_file: app.py | |
| pinned: false | |
| short_description: API for visualizing LLM token probabilities and embeddings | |
| # Token Probability Visualization API | |
| This Space provides API endpoints for visualizing token probabilities and embeddings from language models. | |
| ## Features | |
| - **Token Probabilities**: Get next-token probability distributions for any input text | |
| - **Embeddings**: Access token embeddings with PCA projections for visualization | |
| - **Comparative Analysis**: Compare conditional vs unconditional distributions | |
| - **Multiple Contexts**: Analyze probability shifts across different text contexts | |
| ## API Endpoints | |
| ### `/predict` - Main Prediction Endpoint | |
| Returns comprehensive token probabilities, embeddings, and PCA projections. | |
| ### `/embeddings` - Token Embeddings | |
| Get embeddings for specific token IDs. | |
| ## Usage | |
| Use the API to build interactive visualizations showing: | |
| - Token probability landscapes | |
| - Embedding space visualizations | |
| - Probability distribution comparisons | |
| - Token movement analysis | |
| Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference | |