Hacker News Reader

lawrencechenon Launch HN: Retell AI (YC W24) – Conversational Speech API for Your LLM

thanks!

lawrencechenon Launch HN: Retell AI (YC W24) – Conversational Speech API for Your LLM

Can you buy/remove phone numbers programmatically? I can't find anything in the docs.

lawrencechenon Launch HN: Retell AI (YC W24) – Conversational Speech API for Your LLM

Could you add Partykit/Cloudflare Durable Objects integration docs?

lawrencechenon Sora: Creating video from text

> Historically, letting technology eliminate their jobs has been a sacrifice people have made for their kids' sakes. Not intentionally, for the most part, but their kids ended up with the new jobs created as a result. No one weaves now, and that's fine.

- paulg

https://twitter.com/paulg/status/1757794178548048117

lawrencechenon 20x faster than pgvector: HNSW index in Postgres with pg_embedding

https://www.postgresql.org/about/licence/

lawrencechenon We raised a bunch of money

> There are two kinds of platform companies : the kind where you can get your Python or Rust or Julia code running nicely, and the kind where you find a way to recompile it to Javascript.

Playful jab on Cloudflare and Deno. Would be interesting if Fly took up hosting "WASM containers" in addition to Docker containers.

lawrencechenon Watch GPT-4 Play Minecraft (Voyager)

https://twitter.com/DeveloperHarris/status/16647320832800235...

lawrencechenon Llama.cpp: Port of Facebook's LLaMA model in C/C++, with Apple Silicon support

Georgi just added support for all models (13B/33B/65B) [0]

LLaMA 65B can do ~2 tokens per second on my M1 Max / 64 gb ram [1]

[0] https://twitter.com/ggerganov/status/1634488664150487041 [1] https://twitter.com/lawrencecchen/status/1634507648824676353

lawrencechenon Launch HN: Second (YC W23) – AI bots that add features to web apps

Just don't add background music :)

lawrencechenon [dead]

https://github.com/geohot/tinygrad/tree/llama

lawrencechenon Show HN: I trained an AI model on 120M+ songs from iTunes

Wow, I love this! Can you make clicking play change the url so that songs are added to our browser history? Perhaps make "refreshed" pages idempotent via a hash in the url?

lawrencechenon Bun v0.2.0

child_process is coming soon: https://twitter.com/jarredsumner/status/1579898972650381312

lawrencechenon 62 People own the same as half the world, reveals Oxfam Davos report (2016)

The average person is nevertheless still better off than they were in all of history.

Once basic needs are met (and tbh far exceeded in most first world countries), it seems that awareness about inequality only makes people less happy.

lawrencechenon Plink: Create epic music together with friends and strangers

This multiplayer live coding/composition in lisp is worth checking out: https://youtu.be/v4rcAIVvPQ8?t=3591

(video linked to in gibber.cc's homepage, which features gibber.cc in other segments)

lawrencechenon Lexical – a web text editor framework that powers Facebook

I put Lexical through Hog Bay's Moby Dick test [0] (paste the entirety of Moby Dick, scroll to the middle, and try editing). It didn't pass. AFAIK, only ProseMirror passes, albeit with some lag.

[0] https://www.hogbaysoftware.com/posts/moby-dick-workout/

lawrencechenon Solid.js feels like what I always wanted React to be

The author of Solid.js (Ryan Carniato) also does weekly livestreams about a range of JavaScript topics:

https://www.youtube.com/c/RyanCarniato9

lawrencechenon Solid.js feels like what I always wanted React to be

^ What about Figma? Photoshop? Not everything is only a CRUD app.

lawrencechenon Show HN: Write universally accessible SQL, not library-specific ORM wrapper APIs

Great (and fun!) resource. https://pgexercises.com/ for more in depth on aggregates + recursion.

lawrencechenon Show HN: I'm working on a open-source, self-hosted alternative to Disqus

Neat! I checked out your source code, and if there's one comment I have, it's the way you query for comments. Currently, response times are around 300-400 ms for just 79 bytes of data. That's likely because you recursively fetch nested comments from a serverless function, which potentially has to make multiple network request to talk to the database. Coincidentally, a past Disqus employee has a great write up on the faster solution (recursive CTE): https://cra.mr/2010/05/30/scaling-threaded-comments-on-djang...

Regardless, great work! Choosing Svelte to hit that 4.7kb mark is super sweet!