It's really hard to keep up with open source machine learning. Almost every new Python repo on GitHub is an ML repo, so we made a website that displays all the latest Python repos in a HN-like list. We also added Replicate and HuggingFace models, and posts from r/{LocalLLaMA,MachineLearning,StableDiffusion}.
The website is updated every hour via GitHub Actions.
npm install
cp .dev.vars.example .dev.vars # add your REPLICATE_API_TOKEN
npm run dev # uses local D1 databaseTo develop against the remote D1 database:
wrangler dev --remote# 1. Create D1 database
wrangler d1 create hype
# 2. Update database_id in wrangler.jsonc with the ID from step 1
# 3. Run migration
wrangler d1 execute hype --remote --file=migrations/0001_init.sql
# 4. Set secrets
wrangler secret put REPLICATE_API_TOKEN
# 5. Deploy
npm run deployContent updates hourly via GitHub Actions, which calls the /api/update endpoint.
Manual trigger: npm run updateContent
Check out Replicate. We make it easy to run, push, and scale machine learning models.