Lightnews — Scholar-powered news

Charting and Navigating Hugging Face's Model Atlas

7 15

Nicolas Yax @nicolasyax.bsky.social · Apr 26

Curious about LLM interpretability and understanding ? We borrowed concepts from genetics to map language models, predict their capabilities, and even uncovered surprising insights about their training !

Come see my poster at #ICLR2025 3pm Hall 2B #505 !

4

Nicolas Yax @nicolasyax.bsky.social · Apr 24

If you are interested in this line of research of mapping LLMs you might also want to check the amazing work of Eliahu Horwitz arxiv.org/abs/2503.10633 and Momose Oyama arxiv.org/abs/2502.16173 10/10

As there are now millions of publicly available neural networks, searching and analyzing large model repositories becomes increasingly important. Navigating so many models requires an atlas, but as mo...

Nicolas Yax @nicolasyax.bsky.social · Apr 24

In short, PhyloLM is a cheap and versatile algorithm that generates useful representations for LLMs that can have creative applications in pratice. 9/10
paper : arxiv.org/abs/2404.04671
colab : colab.research.google.com/drive/1agNE5...
code : github.com/Nicolas-Yax/...
ICLR : Saturday 3pm Poster 505

PhyloLM - a Hugging Face Space by nyax

Nicolas Yax @nicolasyax.bsky.social · Apr 24

A PhyloLM collaborative Huggingface space is available to try the algorithm and visualize maps : huggingface.co/spaces/nyax/... The Model Submit button has been temporarily suspended for technical reasons but it should be back very soon ! 8/10

This app allows you to explore and compare language models through various visualizations, including similarity matrices, 2D scatter plots, and tree diagrams. You can search for models by name, adj...

huggingface.co

2 2

Nicolas Yax @nicolasyax.bsky.social · Apr 24

By using code related contexts we can obtain a fairly different map. For example we notice that Qwen and GPT-3.5 have a very different way of coding compared to the other models which was not visible on the reasoning map. 7/10

Nicolas Yax @nicolasyax.bsky.social · Apr 24

The contexts choice is important as it reflects different capabilities of LLMs. Here on a general reasoning type of context we can plot a map of models using UMAP. The larger the edge, the closer models are from each other. Models on the same cluster are even closer ! 6/10

Nicolas Yax @nicolasyax.bsky.social · Apr 24

It can also measure quantization efficiency by observing the behavioral distance between LLM and quantized versions. In the Qwen 1.5 release, GPTQ seems to perform best. This new concept of metric could provide additional insights to quantization efficiency. 5/10

Nicolas Yax @nicolasyax.bsky.social · Apr 24

Aside from plotting trees, PhyloLM similarity matrix is very versatile. For example, running a logistic regression on the distance matrix makes it possible to predict performance of new models even from unseen families with good accuracy. Here is what we got on ARC. 4/10

Nicolas Yax @nicolasyax.bsky.social · Apr 24

Not taking into account these requirements can still produce efficient distance vizualisation trees. However it is important to remember they do not represent evolutionary trees. Feel free to zoom in to see model names. 3/10

Nicolas Yax @nicolasyax.bsky.social · Apr 24

Phylogenetic algorithms often require common ancestors to not appear in the objects studied but are clearly able to retrieve the evolution of the family. Here is an example in the richness of open-access model : @teknium.bsky.social @maximelabonne.bsky.social @mistralai.bsky.social 2/10

MAGELLAN: Metacognitive predictions of learning progress guide...

Nicolas Yax @nicolasyax.bsky.social · Apr 24

We build a distance matrix from comparing outputs of LLMs to a hundred of different contexts and build maps and trees from this distance matrix. Because PhyloLM only requires sampling very few tokens after a very short contexts the algorithm is particularly cheap to run. 1/10

1 3

Nicolas Yax @nicolasyax.bsky.social · Apr 24

🔥Our paper PhyloLM got accepted at ICLR 2025 !🔥
In this work we show how easy it can be to infer relationship between LLMs by constructing trees and to predict their performances and behavior at a very low cost with @stepalminteri.bsky.social and @pyoudeyer.bsky.social ! Here is a brief recap ⬇️

3 5 15

Reposted by Nicolas Yax

cartathomas.bsky.social @cartathomas.bsky.social · Mar 24

🚀 Introducing 🧭MAGELLAN—our new metacognitive framework for LLM agents! It predicts its own learning progress (LP) in vast natural language goal spaces, enabling efficient exploration of complex domains.🌍✨Learn more: 🔗 arxiv.org/abs/2502.07709 #OpenEndedLearning #LLM #RL

Open-ended learning agents must efficiently prioritize goals in vast possibility spaces, focusing on those that maximize learning progress (LP). When such autotelic exploration is achieved by LLM...

Generalization or Memorization: Data Contamination and Trustworthy Evaluation for Large Language Models

1 3 9

Reposted by Nicolas Yax

cedric @ccolas.bsky.social · Nov 27

we are recruiting interns for a few projects with @pyoudeyer
in bordeaux
> studying llm-mediated cultural evolution with @nisioti_eleni
@Jeremy__Perez

> balancing exploration and exploitation with autotelic rl with @ClementRomac

details and links in 🧵
please share!

1 6 6

Reposted by Nicolas Yax

Gautier Hamon @hamongautier.bsky.social · Nov 22

Putting some Flow Lenia here too

1 1 4

Reposted by Nicolas Yax

Gautier Hamon @hamongautier.bsky.social · Nov 22

1/⚡️Looking for a fast and simple Transformer baseline for your RL environment in JAX ?
Sharing my implementation of transformerXL-PPO: github.com/Reytuag/tran...
The implementation is the first to attain the 3rd floor and obtain advanced achievements in the challenging Craftax

1 1 3

Nicolas Yax @nicolasyax.bsky.social · Nov 15

Related work on contamination in LLMs :
arxiv.org/abs/2402.15938 Dong et al. 2024
arxiv.org/abs/2310.15007 Meeus et al. 2023
arxiv.org/abs/2310.17623 Oren et al. 2024

Recent statements about the impressive capabilities of large language models (LLMs) are usually supported by evaluating on open-access benchmarks. Considering the vast size and wide-ranging sources of...

Assessing Contamination in Large Language Models: Introducing the LogProber method

Nicolas Yax @nicolasyax.bsky.social · Nov 15

LogProber paper : www.arxiv.org/abs/2408.14352
git : github.com/Nicolas-Yax/...
collab : colab.research.google.com/drive/1GDbmE...

In machine learning, contamination refers to situations where testing data leak into the training set. The issue is particularly relevant for the evaluation of the performance of Large Language Models...

www.arxiv.org

Nicolas Yax @nicolasyax.bsky.social · Nov 15

It is part of a research agenda to open the LLM black box and provide tools for researchers to better interact with models in a more transparent manner. The last paper in this agenda was PhyloLM proposing methods to investigate the phylogeny of LLMs arxiv.org/abs/2404.04671 15/15

Nicolas Yax @nicolasyax.bsky.social · Nov 15

This method was first introduced in our paper Studying and improving reasoning in humans and machines investigating the evolution of cognitive biases in language models. www.nature.com/articles/s44... 14/15

www.nature.com

1 1

Nicolas Yax @nicolasyax.bsky.social · Nov 15

As such, LogProber can be a useful tool to check contamination in language models at a very low cost (one forward pass) given some very high level assumptions about the training method (that are very often verified in practice). 13/15

Nicolas Yax @nicolasyax.bsky.social · Nov 15

Lastly, the A scenario is more common in instruction finetuning scenarios. In open access models finetuning databases are often shared making it possible to check directly if the item is found in the training set which is rarely the case for pretraining databases. 12/15