Spaces:
Runtime error
Runtime error
File size: 1,834 Bytes
5b76cb0 e8e08ec 5b76cb0 9cdcce8 5b76cb0 3839eee 5b76cb0 7b406f4 fb59416 5b76cb0 9cdcce8 5b76cb0 3839eee 5b76cb0 7b406f4 fb59416 5b76cb0 9cdcce8 5b76cb0 9cdcce8 9d7051c ab74a45 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 | # Deployment
## Hugging Face Space
- Space repo: `https://huggingface.co/spaces/KPrashanth/agent-tina`
- Live app: `https://kprashanth-agent-tina.hf.space/`
- SDK: Docker
- App port: `7860`
- Runtime status last verified: `RUNNING` on 2026-06-08
- Dataset repo: `https://huggingface.co/datasets/KPrashanth/agent-tina-meetings`
- End-to-end meeting creation, transcription, correction, MoM generation, downloads, and Dataset persistence verified on 2026-06-08
## Local Deployment Script
Run:
```powershell
.\.venv\Scripts\python.exe scripts\deploy_hf_space.py
```
The script:
- reads credentials from `.env`
- creates or updates the Space
- uploads project files
- ignores `.env`, `.venv`, caches, bytecode, and `sample.wav`
- configures Space variables/secrets
## Required `.env` Values
```powershell
HF_TOKEN=...
HF_SPACE_NAME=agent-tina
HF_APP_BASE_URL=https://kprashanth-agent-tina.hf.space
HF_DATASET_REPO=KPrashanth/agent-tina-meetings
OPENROUTER_API_KEY=...
OPENROUTER_MODEL=openai/gpt-oss-20b
OPENROUTER_CORRECTION_MODEL=openai/gpt-4o-mini
WHISPER_MODEL=base
WHISPER_BEAM_SIZE=5
MAX_UPLOAD_MB=250
```
## Space Secrets and Variables
Secret:
- `OPENROUTER_API_KEY`
- `HF_TOKEN`
Variables:
- `OPENROUTER_MODEL`
- `OPENROUTER_CORRECTION_MODEL`
- `WHISPER_MODEL`
- `WHISPER_BEAM_SIZE`
- `MAX_UPLOAD_MB`
- `APP_BASE_URL`
- `HF_DATASET_REPO`
## Important Constraint
Hugging Face Space disk may be ephemeral unless persistent storage is enabled. The app optionally syncs meeting folders to `HF_DATASET_REPO` when `HF_TOKEN` and `HF_DATASET_REPO` are configured.
After a Space rebuild, meeting metadata is restored from the private Dataset on demand. Large original recordings and Markdown outputs are restored lazily only when requested.
Post-rebuild metadata and original-audio recovery were verified on 2026-06-09.
|