jcuenod

7 ms·

31.
by jcuenod 2y ago
Lmsys is not a measure of intelligence. It's a measure of human preference. People prefer correct answers (assuming they are qualified to identify the correct one), but they also prefer answers formatted nicely for reading, for example
32.
by jcuenod 2y ago
The problem is that Svelte 5 is not Javascript in the worst way. It doesn't solve the problems of js, it provides leaky abstractions around some frontend issues. I think Solid is a better direction than Svelte because Solid doesn&#x2
33.
Beyond TypeScript(jcuenod.github.io)
1 points by jcuenod 2y ago|0 comments
34.
by jcuenod 2y ago
it will be exactly what I want!
35.
Ask HN: A Better TypeScript?
1 points by jcuenod 2y ago|2 comments
36.
by jcuenod 2y ago
This looks lovely! Good work. On my wishlist are some fuzzier categories: 1. Tech trends (rust, docker, postgres...) 2. Role trends ("ml engineer", "full-stack developer" ...)
37.
Show HN: Mammal – (Yet another) LLM client built on Tauri(github.com)
1 points by jcuenod 2y ago|0 comments
38.
Advancing the Accuracy-Efficiency Frontier with Llama-3.1-Nemotron-51B(developer.nvidia.com)
1 points by jcuenod 2y ago|0 comments
39.
by jcuenod 2y ago
Seems like setting max_tokens crashes your endpoint.
40.
by jcuenod 2y ago
It warms the cockles of my heart to see this kind of thing. Happened to me recently with a 24 year old firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=62151
41.
by jcuenod 3y ago
This deserves its own Show HN!
42.
StarCoder2 and the Stack v2(huggingface.co)
1 points by jcuenod 3y ago|0 comments
43.
by jcuenod 3y ago
And for those who are interested, Evan (creator of Elm) gave a great Strange Loop '23 talk that hints at where he imagines things going, if he could figure out how to get the economics to work. https://www.youtube.com/w
44.
by jcuenod 3y ago
This looks more promising to me than what I saw yesterday. There is an "Open" aspect to their objectives and they do say: > We’re committed to empower the AI community with open technology. Our open models sets the bar for effi
45.
by jcuenod 3y ago
Can't tell if you're responding to me or parent. But that's what I was gonna say to @canjobear...
46.
by jcuenod 3y ago
Came here to post the same thing for Phi-2: +-------------+----------+-------------+ | Benchmark | Gemma 2B | Phi-2 2.7B | +-------------+----------+-------------+ | MMLU | 42.3 | 56.7 | | MBPP | 29
47.
by jcuenod 3y ago
Then they're not "emergent"
48.
by jcuenod 3y ago
It won't remove the use of RAG at all. That's like saying, "wow, now that I've upgraded my 128GB HDD to 1TB, I'll never run out of space again."
49.
by jcuenod 3y ago
It's the dev cycle: "Hey look pure functional components... and now they also have state... and side effects!"
50.
BASE TTS: Lessons from building a billion-parameter Text-to-Speech model(arxiv.org)
3 points by jcuenod 3y ago|0 comments
51.
BASE TTS: The largest text-to-speech model to-date(amazon-ltts-paper.com)
200 points by jcuenod 3y ago|78 comments
52.
by jcuenod 3y ago
Exactly this. If you have a model that's never seen JSON and you want JSON to come out, fine-tuning probably not a bad idea. If you have a model trained on English documents and you want it to produce English documents related to your
53.
by jcuenod 3y ago
Trying asking an LLM :)
54.
by jcuenod 3y ago
Thanks! I haven't heard of these, but I'll take a look. And maybe that's the trick, alcohol during the day would curb my enthusiasm for pouring my drinks all over my desk.
55.
Ask HN: Keyboard/Mouse Recommendations
1 points by jcuenod 3y ago|2 comments
56.
by jcuenod 3y ago
It's with respect to the smell
57.
by jcuenod 3y ago
For real this time?
58.
by jcuenod 3y ago
Depends whether the cat is dead or alive
59.
by jcuenod 3y ago
Harlequin has been on my "to investigate" list since it popped up on hn a few weeks/months ago. I still need to experiment with it a bit more, but I had been thinking of building a some tragic equivalent to this, and now I do
60.
by jcuenod 3y ago
From their FAQs: We are a small team, so it's critical for us to be laser-focused. As a startup, one of our key priorities at this early phase is learning, and right now, we're focused on the following questions: What are

More ›