HNHacker News
TopNewBestAskShowJobs

xfalcox

1,042 karma · joined November 7, 2013

[ my public key: https://keybase.io/falcofantastic; my proof: https://keybase.io/falcofantastic/sigs/_PKYsKf2wmCyt834lEh6N4POje9RoICd3Ta7qezTzJE ]
submissionscomments
xfalcox··on Show HN: A backend agnostic Ruby framework for building reactive desktop apps
This looks super cool, exactly what I've been wanting to create some useful widgets! Thanks for sharing!
xfalcox··on I analyzed chord progressions in 680k songs
I guess one aspect missing here is weighting more popular songs on that analysis.

I assume that the analysis is simply counting every song chords, so a unknown band you've never heard about has the same impact as The Ramones.

I'd like to see the same graph weighted by band popularity using either YouTube or Spotify data.

xfalcox··on Cohere Launches Embed 4
No downloadable open weights ?

Looks like I'll stay on [bge-m3](https://huggingface.co/BAAI/bge-m3)

xfalcox··on Brazil's government-run payments system has become dominant
My wife runs a small retail makeup shop on Shopify, which started before pix and those surprise false fraud claims almost killed the business.

Pix was such a game changer. It is perfect.

xfalcox··on Foreign visits into the U.S. fell off a cliff in March
I just got an invite to Google I/O, which I've always dreamed of attending, but as someone from South America I'm having nightmares about ICE.

It's just a 4 days trip, but I have an entire family that needs me to come back and not sent for a decade to a random overseas prison.

xfalcox··on Show HN: I built a(nother) house optimized for LAN parties
This is super awesome, congrats!
xfalcox··on Binary vector embeddings are so cool
I shipped this in Discourse a couple of weeks ago, and it's incredible tech. And fully supported in PostgreSQL via pgvector.
xfalcox··on Understanding Ruby 3.3 Concurrency: A Comprehensive Guide
Can I safely use Ractors inside Rails if my tasks have a mix of CPU and I/O ?
xfalcox··on Web Assembly audio decoders highly optimized for size and performance
I wish the author would do the same for image decoding. iOS inability to draw a canvas with 4096x4097 dimensions means you can't use an Off screen canvas to go from file to RGBA array.
xfalcox··on Initial CUDA Performance Lessons
What about FP8? It is a target that is very popular for LLM inference.
xfalcox··on COSMIC Alpha Released
I've been daily driving it on arch since the day it released and it's quite usable, with minimal bugs for my needs.

I just wish they add an applet that integrates Google Calendar stuff in their calendar, so I can know if I have a meeting or something coming up at a glance.

xfalcox··on Kagi LLM Benchmarking Project
Is it just me or the 2 Llama 3.1 providers are the two know for running quantization levels that undermine Llama perf ?
xfalcox··on Show HN: I created an After Effects alternative
Can I just say that I was disappointed by the Web Codecs API leaving muxing the end result as a "draw the rest of the fucking owl" thing.
xfalcox··on Claude 3.5 Sonnet
I still get

{"message":"Could not resolve the foundation model from the provided model identifier."}

on us-west-2.

xfalcox··on Claude 3.5 Sonnet
They said it is available on Bedrock, but it isn't :(
xfalcox··on HuggingFace text-generation-inference is reverting to Apache 2.0 License
Worth noting that this also impacts the great https://github.com/huggingface/text-embeddings-inference, which allows anyone to run state of the art embeddings with great performance.
xfalcox··on Command R+: A Scalable LLM Built for Business
Their other license also prohibits that

> Cohere For AI Acceptable Use Policy > generating synthetic data outputs for commercial purposes, including to train, improve, benchmark, enhance or otherwise develop model derivatives, or any products or services in connection with the foregoing.

xfalcox··on Jpegli: A new JPEG coding library
Does anyone have compiled this to WASM? I'm currently using MozJPEG via WASM for a project and would love to test replacing it by Jpegli.
xfalcox··on I scraped all of OpenAI's Community Forum
> Is that within the defined data processing purposes of all Discourse setups?

It's an optional plugin that can be enabled / disabled by the site admin. Those modules are all disabled by default, and each need to be enabled by the site owner.

> Edit: and apparently it also tries to flag NSFW chat messages, does Discourse have PM chats where this would flag private messages for admins to read or is it only public chats that this bot runs on?

Discourse PMs can be read by admins, see the definition here: https://meta.discourse.org/t/guidance-and-best-practices-on-...

xfalcox··on I scraped all of OpenAI's Community Forum
That's super cool, thanks for sharing! I will share this as an easy to follow example of what we can with AI.

> Allowing a Q&A interface using these embeddings over the post contents could speed up research over the community posts (if you know the right questions to ask :P). Let's view some posts similar to this one complaining about function calling

That's indeed a great thing to surface, and that's exactly how the the OpenAI forum selects the "Related Topics" to show at the end of every topic. We use embeddings for this feature, and the entire thing is open-source: https://github.com/discourse/discourse-ai/blob/main/lib/embe...

We also embeddings for suggesting tags, categories, HyDE search and more. It's by far my favorite tech of this new AI/ML gen so far in terms of applicability.

> Using Twitter-roBERTa-base for sentiment analysis, we generated a post_sentiment label (negative, positive, neutral) and post_sentiment_score confidence score for each post.

We do the same, with even the same model, and conveniently show that information on the admin interface of the forum. Again all open source: https://github.com/discourse/discourse-ai/tree/main/lib/sent...

Disclaimer: I'm the tech lead on the AI parts of Discourse, the open source software that powers OpenAI's community forum.

xfalcox··on YouTube video embedding harm reduction
Adding a lazy, or going the extra mile of only loading a pure HTML div that is swapped by the iframe on click would be better to limit tracking. Do not load the iframe for people who don't want to play it.
xfalcox··on Understanding, using, and finetuning Gemma
> Gemma 7B is a 9B model. The name is a lie

Ohhh so that explains why I couldn't load it on my RTX 4090, while other 7B models load just fine!

xfalcox··on Show HN: Natural-SQL-7B, a strong text-to-SQL model
Context is 4096? My app db DDL is 19877 tokens (using Llama2 tokenizer) long, so that means we need to do a RAG for handling the DDL prompt injection.

A model like this with a 32k long seq_len, like Mixtral, would be a killer for me.

xfalcox··on Show HN: Open-source Rule-based PDF parser for RAG
We've been looking for something exactly like this, thanks for sharing!
xfalcox··on Posthog is closing their Slack community in favor of forum
There are many plugins, check https://meta.discourse.org/c/plugin/22

The flexibility of Ruby makes it kinda easy to change behavior.

Full Disclosure, I work there.

xfalcox··on Features and Benefits of Bluetooth Mesh 1.1 for Wireless Mesh Networking
Can you share some of those ZigBee products? I'm looking into getting started on that, but Costa has been holding me back.
xfalcox··on Gemini AI
I'm researching using LLMs for alt-text suggestion for forum users, can you share your finding so far?

Outside of GPT-4V I had good first results with https://github.com/THUDM/CogVLM

xfalcox··on Workers AI: Serverless GPU-powered inference
Embedding cost and model choice makes this a very compelling choice. I'm working on leveraging embeddings in https://github.com/discourse/discourse-ai where it powers offering related topics, semantic search, tag and category recommendations among other things.

A cheap offering like this can make it a lot more reasonable for self-hosters.

xfalcox··on Firefox 118
I wonder what the deal with Google Meet was. Did Firefox fix this? Did Google team fix this? If this is retroactively fixed for the last 3 versions it must be a google-side fix right?
xfalcox··on Show HN: Dataherald AI – Natural Language to SQL Engine
We are experimenting in this area, as a way to provide a way for community managers to query their Discourse PG database using natural language.

Your product is very interesting. Maybe we can collaborate here, as Discourse is a standard Rails database schema on PostgreSQL and could be a good way to show your tool working in a real database.

← PreviousPage 2 of 11Next →