README
nostr-notes-labeler
A Python service that consumes notes from a Nostr relay and enriches them with machine-learning labels.
It connects to a relay (default wss://saltivka.org) over WebSocket, streams in events, and for each note runs:
- Language identification — Facebook's fastText language ID model, mapped to ISO 639-2 codes
- Sentiment analysis — a 🤗 Transformers sentiment pipeline
- Named-entity recognition —
dslim/bert-base-NERto extract entities (people, organisations, locations)
Redis is used for coordination/state. Models run on CPU by default, or on GPU when CUDA_DEVICE is set.
Usage
With Docker Compose:
docker compose up
Or directly:
pip install -r requirements.txt
python3 app.py
Configuration
All configuration is via environment variables:
| Variable | Default | Description |
|---|---|---|
RELAY_URL | wss://saltivka.org | Nostr relay to consume from |
REDIS_URL | redis://redis:6379 | Redis connection URL |
SECRET_KEY | — | Secret key used by the service (provide your own; do not commit it) |
PAST_EVENTS_LIMIT | 1000 | How many historical events to pull on connect |
CUDA_DEVICE | -1 | GPU device index; -1 runs the models on CPU |
Contributors
2 commits
Similar repos
Languages
Python
95.5%
Dockerfile
4.5%
README
nostr-notes-labeler
A Python service that consumes notes from a Nostr relay and enriches them with machine-learning labels.
It connects to a relay (default wss://saltivka.org) over WebSocket, streams in events, and for each note runs:
- Language identification — Facebook's fastText language ID model, mapped to ISO 639-2 codes
- Sentiment analysis — a 🤗 Transformers sentiment pipeline
- Named-entity recognition —
dslim/bert-base-NERto extract entities (people, organisations, locations)
Redis is used for coordination/state. Models run on CPU by default, or on GPU when CUDA_DEVICE is set.
Usage
With Docker Compose:
docker compose up
Or directly:
pip install -r requirements.txt
python3 app.py
Configuration
All configuration is via environment variables:
| Variable | Default | Description |
|---|---|---|
RELAY_URL | wss://saltivka.org | Nostr relay to consume from |
REDIS_URL | redis://redis:6379 | Redis connection URL |
SECRET_KEY | — | Secret key used by the service (provide your own; do not commit it) |
PAST_EVENTS_LIMIT | 1000 | How many historical events to pull on connect |
CUDA_DEVICE | -1 | GPU device index; -1 runs the models on CPU |
Similar repos
Contributors
2 commits
Languages
Python
95.5%
Dockerfile
4.5%