lucrbvi

45 ms·

1.
by lucrbvi 4d ago
> This is why it seems that alignment for a generalized, chat-style AI is a very hard problem, perhaps impossible. You can't align it to solve a certain kind of problem and keep it general to any question. The two goals are in confl
2.
by lucrbvi 5d ago
That's a lot of ®, curious how ClickHouse® Inc. is treating the use of its name by others ... Hopes it's not like Oracle with JavaScript
3.
by lucrbvi 9d ago
The target model is the original LLM that is large and expensive. It can verify candidate tokens in a single forward pass. It means you give all the context + candidate tokens that passes in parallel in the backbone, then you pass the langu
4.
by lucrbvi 21d ago
Under the EU's AI Act non-profits and universities can just bypass these constents to make non-commercial datasets btw
5.
by lucrbvi 1mo ago
It would be fun to train a small vision neural network to recognize ads instead of relying on HTML or network requests. I don't know if anyone has done this yet.
6.
by lucrbvi 1mo ago
https://www.pangram.com/history/c2abdcf5-c52b-41cc-a8fb-0c59...
7.
by lucrbvi 1mo ago
Mistral has the capability of training such models. Take a look at Poolside[1], they are claiming to pre-train their Laguna series of models on 4,096 NVIDIA H200 GPUs[2]. Mistral has approximately 13,800 NVIDIA GB300 GPUs, which are nearly
8.
by lucrbvi 2mo ago
"world model" is, sadly, a new buzzword; most of the world models we are seeing are video generative models (like GPT-Image), they receive an action vector and previous frames and must generate the next frame. It's working of
9.
by lucrbvi 2mo ago
Seems like MAI models from Microsoft are not going to be open-weight soon, but they are sharing a lot of details in the making of these models, which is a weird position.
10.
by lucrbvi 2mo ago
Such a gem, thanks to the author for sharing it's findings :) The only problem I have with planing in latent space is that it can be really noisy and not representative of the positions in the game (the latent are trained for semantic,
11.
by lucrbvi 2mo ago
Anthropic theorize that middle layers in an LLM is a "J-Space" used to "think" about the future answer or about abstract concepts. Their method is used to identify which tokens can appears in which layers of the model.
12.
by lucrbvi 3mo ago
http://archive.today/1YRCE
13.
by lucrbvi 3mo ago
https://xkcd.com/927/
14.
Building a Latent World Model of Doom(lucrbvi.com)
1 points by lucrbvi 4mo ago|0 comments
15.
by lucrbvi 4mo ago
In the article the author says they are doing reinforcement learning with LLMs.
16.
by lucrbvi 4mo ago
How does this differ from dottxt's Outlines[0] on the technical level? Are you using some JSON grammar to force the LM head distribution to follow it? [0]: https://github.com/dottxt-ai/outlines
17.
by lucrbvi 4mo ago
it's using webgpu kernels, probably a false positive
18.
by lucrbvi 4mo ago
I really like QBE but I hope they will make it a true library because I don't want to launch a subprocess to compile a program.
19.
by lucrbvi 4mo ago
Cool but why would anyone write an agent in C? Since the language is verbose it seems to be a painful choice to use C for this task.
20.
by lucrbvi 5mo ago
They have added a lot of optimization focussing on the KV-cache, so they can have a much larger window without eating all the VRAM. The 1M window might be usable, but it will probably underperform against a smaller window of course.
21.
World Models will push the frontier for LLMs(lucrbvi.bearblog.dev)
3 points by lucrbvi 5mo ago|0 comments
22.
by lucrbvi 5mo ago
Mollie seems to only provide services to business based in European Economic Area, Switzerland and the UK [0], so yes? [0]: https://help.mollie.com/hc/en-us/articles/115002116105-Can-I...
23.
by lucrbvi 5mo ago
Mollie might be a direct competitor
24.
by lucrbvi 5mo ago
Is there anyone that feels that LLMs are wrong for computer use? It's like robotic, if find LLMs alone are really slow for this task
25.
by lucrbvi 5mo ago
Some people are speculating that Opus 4.7 is distilled from Mythos due to the new tokenizer (it means Opus 4.7 is a new base model, not just an improved Opus 4.6)
26.
Why Claude Mythos is not AGI(lucrbvi.bearblog.dev)
2 points by lucrbvi 5mo ago|1 comments
27.
by lucrbvi 5mo ago
Hi! I share my first blog post ever about the definition on AGI following the annoucement of Claude Myhtos Preview. I hope some of you might found this interesting.
28.
by lucrbvi 6mo ago
Sounds like Multi-Head Latent Attention (MLA) from DeepSeek
29.
by lucrbvi 6mo ago
I wouldn't be surprised if Vercel were bought by Anthropic/OAI (but maybe it would be too expensive?)
30.
by lucrbvi 6mo ago
This is a weird pattern accross OpenAI/Anthropic to buy startups building better toolings. I don't really see the value for OAI/Anthropic, but it's nice to know that uv (+ ty and many others) and Bun will stay maintained

More ›