lewtun

11 ms·

1.
by lewtun 5mo ago
Shameless plug: https://huggingface.co/spaces/smolagents/ml-intern It’s a simple harness around Opus, but with tight integration to Hugging Face infra, so the agent can read papers, test code and launch experiment
2.
by lewtun 6mo ago
Hugging Face Buckets are pretty simple: https://huggingface.co/docs/huggingface_hub/en/guides/bucket... Disclaimer: I work at HF
3.
by lewtun 11mo ago
The analogy stems from the notion that neural nets are "grown" rather than "engineered". Chris Olah has an old, but good post with some specific examples: https://colah.github.io/notes/bio-analogies&
4.
by lewtun 11mo ago
Thanks! I expect the book will remain relevant as long as the Transformers architecture does. That’s why we mostly focus on topics we think will stand the test of time, but let’s see how that plays out :)
5.
by lewtun 11mo ago
In the specific case of SmolLM, it originates from the meme in this dataset https://huggingface.co/datasets/bigcode/the-stack-smol
6.
by lewtun 11mo ago
Hi, Lewis here (one of the co-authors). Happy to answer any questions people have about the book :)
7.
PyTorch OpenEnv(github.com)
1 points by lewtun 1y ago|0 comments
8.
Scaling Laws for Reinforcement Learning(huggingface.co)
1 points by lewtun 1y ago|0 comments
9.
by lewtun 1y ago
For those interested in playing with an implementation of these ideas, my colleagues at HF made some recipes here: https://github.com/huggingface/trl/blob/main/docs/source/lor...
10.
by lewtun 1y ago
“QED and the Men Who Made It” [1] might be close to what you’re after for quantum theory at least. Unlike other popular accounts, it gets quite technical and covers a lot of the historical dead ends that people had during the development of
11.
by lewtun 1y ago
> We instantiate this idea through Preference-prior Informed Linucb fOr adaptive rouTing (PILOT), a novel extension of LinUCB Academics are pretty creative at naming their creations
12.
by lewtun 1y ago
Indeed we opted for offline methods like Anchored Preference Optimization as we found in the Open R1 project that doing multi-task RL on small models is quite a hassle to get right. With offline methods, you focus much more on dataset curat
13.
by lewtun 1y ago
> The absolute best way of doing this is these days is likely through a vision based machine learning model, but that is an approach that is very far away from scaling to processing hundreds of gigabytes of PDF files off a single server
14.
DESI results show dark energy may be evolving over time(newscenter.lbl.gov)
1 points by lewtun 2y ago|0 comments
15.
DocumentAI with 256M Parameters(huggingface.co)
5 points by lewtun 2y ago|0 comments
16.
220k reasoning traces from DeepSeek-R1(huggingface.co)
1 points by lewtun 2y ago|0 comments
17.
by lewtun 2y ago
I gave the demo a spin and it’s pretty nice! One thing I noticed is that the avatar doesn’t seem to be aware of it’s surroundings- for example, I asked it why it was wearing a cowboy hat and it was adamant that it wasn’t wearing a hat at al
18.
by lewtun 2y ago
> I expect language models to also get crazy good at mathematical theorem proving Indeed, systems like AlphaProof / AlphaGeometry are already able to win a silver medal at the IMO, and the former relies on Lean for theorem verificat
19.
The largest math dataset of Olympiad problems for training LLMs(huggingface.co)
3 points by lewtun 2y ago|0 comments
20.
by lewtun 2y ago
Hello everyone, we just did a speed run with Argilla and KAIST AI to fine-tune the beefy new Mixtral model with some new techniques that came out recently. More details in the model card - enjoy!
21.
Recipes to align LLMs with AI feedback(github.com)
1 points by lewtun 3y ago|0 comments
22.
A Million AI Preferences(huggingface.co)
1 points by lewtun 3y ago|0 comments
23.
by lewtun 3y ago
The myth is also promoted in Chapter 3 of The Making of the Atomic Bomb by Richard Rhodes: > Plank had taught at Berlin since 1889. In 1900 he had proposed a revolutionary idea to explain a persistent problem in mechanical physics, the s
24.
Constitutional AI with Open LLMs(huggingface.co)
2 points by lewtun 3y ago|0 comments
25.
Zephyr 7B(huggingface.co)
4 points by lewtun 3y ago|0 comments
26.
Diffusion Models Live Event with Hugging Face(huggingface.co)
1 points by lewtun 4y ago|1 comments
27.
by lewtun 4y ago
Talks and discussion with the creators of Stable Diffusion and more :)
28.
by lewtun 8y ago
Spoud | Data Scientist / Senior Software Engineer / DevOps Engineer | Bern, Switzerland | ONSITE | http://spoud.io/jobs.html Spoud is a Series A funded startup that is building a data market platform to enable ent