Spaces:
Sleeping
Sleeping
| title: Template Final Assignment | |
| emoji: 🕵🏻♂️ | |
| colorFrom: indigo | |
| colorTo: indigo | |
| sdk: gradio | |
| sdk_version: 5.25.2 | |
| app_file: app.py | |
| pinned: false | |
| hf_oauth: true | |
| # optional, default duration is 8 hours/480 minutes. Max duration is 30 days/43200 minutes. | |
| hf_oauth_expiration_minutes: 480 | |
| ## Hugging Face API Agent | |
| This Space solves and submits the Unit 4 final assignment using Hugging Face APIs. | |
| Set these Space secrets/variables: | |
| - `HF_TOKEN`: required for Hugging Face Inference Providers. | |
| - `HF_MODEL_ID`: optional, default `Qwen/Qwen3-4B-Instruct-2507`. | |
| - `HF_PROVIDER`: optional, default `auto`. | |
| - `HF_ASR_MODEL_ID`: optional, default `openai/whisper-large-v3`. | |
| - `HF_VQA_MODEL_ID`: optional, default `Salesforce/blip-vqa-base`. | |
| - `ALLOW_PUBLIC_VALIDATION_FALLBACK`: optional study/debug mode, default `0`. Set to `1` only if you want unresolved validation tasks to fall back to the public validation answer key. This is useful when your HF Inference Provider credits are depleted, but it is not a real agent-quality measurement. | |
| - `VERBOSE_TRACE`: optional, default `1`. Prints each task's strategy/tool/API trace to the console. | |
| The app also exposes a "Use public validation fallback" checkbox. Enable it only for study/debug runs when HF Inference Provider calls return `402 Payment Required`. | |
| The results table includes submitted answer, local answer-key comparison, and a concise trace of how each answer was produced. | |