ebalit

8 ms·

1.
by ebalit 1y ago
It seems like wero [0] is making some progress on this front without relying digital euro or on crypto. 0: https://en.wikipedia.org/wiki/Wero_(payment)
2.
Self Forcing: Real-time streaming video generation on a single RTX 4090(self-forcing.github.io)
4 points by ebalit 1y ago|0 comments
3.
by ebalit 1y ago
It's most probably ZenStack given the description: https://zenstack.dev/
4.
Fixing the Ransac Stopping Criterion(arxiv.org)
3 points by ebalit 2y ago|0 comments
5.
by ebalit 2y ago
The analogy between projects/companies and governments is missing big components though. - "Benevolent Dictators" of companies or projects have to obey the law - They can't forbid competition or alternatives - Every part
6.
by ebalit 2y ago
Do you have a chat example you can share where you found it to work great? I feel that the performance varies a lot between domains.
7.
by ebalit 2y ago
You mentioned a model released this week. Is it Lumina new model by any chance?
8.
by ebalit 2y ago
Also commits. I imagine that there is a lot of information to gather from the history of repos in addition to the "static view" of a codebase. However, it doesn't seem trivial to do deduplication in that case without removing
9.
Predicted outputs: GPT-4o inference speed-up for editing tasks(platform.openai.com)
3 points by ebalit 2y ago|0 comments
10.
AMD Open-Source 1B OLMo Language Models(amd.com)
78 points by ebalit 2y ago|33 comments
11.
by ebalit 2y ago
I'm totally with you on the evolution of motor tech because of drone and also personal mobility (scooters and hoverboard motors are a steal for what they can handle). While high torque motors got way cheaper, especially with MIT Cheeta
12.
by ebalit 2y ago
I can't find any information that show a difference as large as 2x. Do you have a specific comparison point in mind? From Nvidia and AMD, I read sparse fp8 at 7 PFLOPs for B100 [0] vs 5.22 PFLOPs for mi325x [1] Nvidia doesn't give
13.
by ebalit 2y ago
But Meta is the main company behind Pytorch development. If they make it work and upstream it, this will cascade to all Pytorch users. We don't have to imagine far, it's slowly happening. Pytorch for ROCm is getting better and bet
14.
by ebalit 2y ago
When you already know what you building, but not when you're doing exploratory data analysis for example. There is a good reason why the ML community took Python as the favorite language overall.
15.
INT8 FlashAttention(arxiv.org)
2 points by ebalit 2y ago|0 comments
16.
by ebalit 2y ago
Hi, Exxa's CTO here. Feel free to ask me anything!
17.
by ebalit 2y ago
I think this might be key, in addition to some landmark tokens to quickly backtrack to. The big question is how to train such model. There is a recent paper from Meta that propose a way to train a model to backtrack its generation to improv
18.
by ebalit 2y ago
Transformers can also fetch at any moment any previous information that become useful . RNN are constantly updating and overwriting their memory. It means they need to be able to predict what is going to be useful in order to store it for
19.
by ebalit 2y ago
There was a HackadayPrize 2023 competitor that worked on this [0]. He had to rethink the way those devices are built to bring the cost down. That would be interesting to know if his solution could match the 4k$ in term of usability or if th
20.
by ebalit 2y ago
He didn't say that Spanner is only a recruitment tool but that the blog posts about Spanner (and other core technologies of Google) might be.
21.
by ebalit 2y ago
I'm not a lawyer but from what I understand the process is closer to what you mention for the US. We have "garde à vue" and "détention provisoire": - "Garde a vue" is similar to being in police custody and
22.
by ebalit 2y ago
We built the cheapest Llama 3.1 70B inference API, specialized for tasks that are not time sensitive (ie. batch processing jobs for example). Without any quantization our current price is 30cts ingest and 50cts output per million tokens. [1
23.
by ebalit 2y ago
My question was genuine as I'm not from the gamedev domain and I might have missed the real state of the art. Hard coded dialogs often feel very unnatural and limiting. I can see why people want to explore LLM to try to make new experi
24.
by ebalit 2y ago
As a gamedev, what would you use currently instead for NPCs? And what would you say makes this/those solutions better than LLMs?
25.
by ebalit 2y ago
But the price should be a factor. Your fair comparison would match a ~60k$ setup to a 20k$ according to prices we can find online. I don't think it should be ignored, especially when the power consumption is similar.
26.
by ebalit 2y ago
You need 2 H100 to have enough VRAM for the model whereas you need only 1 MI300X. Doubling the total throughput (for all completions) of 1 MI300X to simulate the numbers for a duplicated system is reasonable. They should probably show separ
27.
by ebalit 2y ago
vLLM inference of Mixtral in fp16 is a real workload. I guess the details are there because of the different inference engine used. You need the most similar compute tasks to be ran but the compute kernels can't be the same as in the e
28.
by ebalit 2y ago
Very cool! It reminds me of Prezi! https://prezi.com I did an old experiment on a scrollable whiteboard with replay that I built after watching a khan academy style video and wanting to scroll to back to a formula without pausin
29.
PyTorch Library for Running LLM on Intel CPU and GPU(github.com)
308 points by ebalit 2y ago|95 comments
30.
by ebalit 3y ago
Comment on HN are often very negative about AMD / ROCm and the slow release of an offer competitive with Nvidia. But I'm very hopeful. They're slowly chipping away at Nvidia massive lead on the software front. And now everyth

More ›