CompVis

university

CompVis

AI & ML interests

Generative approaches for visual synthesis, Invertible deep models for explainable AI, Deep metric and representation learning, self-supervised learning paradigms

Recent Activity

YfZ authored a paper 15 days ago

Towards Aligned Layout Generation via Diffusion Model with Aesthetic Constraints

YfZ authored a paper 15 days ago

LLaVA-Read: Enhancing Reading Ability of Multimodal Language Models

YfZ authored a paper 15 days ago

ARTIST: Improving the Generation of Text-rich Images by Disentanglement

View all activity

CompVis's activity

akhaliq

posted an update 15 days ago

Post

3471

Google drops Gemini 2.0 Flash Thinking

a new experimental model that unlocks stronger reasoning capabilities and shows its thoughts. The model plans (with thoughts visible), can solve complex problems with Flash speeds, and more

now available in anychat, try it out: akhaliq/anychat

YfZ

authored 7 papers 15 days ago

Towards Aligned Layout Generation via Diffusion Model with Aesthetic Constraints

Paper • 2402.04754 • Published Feb 7, 2024

LoRA-Contextualizing Adaptation of Large Multimodal Models for Long Document Understanding

Paper • 2411.01106 • Published Nov 2, 2024 • 4

SUGAR: Subject-Driven Video Customization in a Zero-Shot Manner

Paper • 2412.10533 • Published 21 days ago • 5

anton-l

posted an update 16 days ago

Post

2104

Introducing 📐𝐅𝐢𝐧𝐞𝐌𝐚𝐭𝐡: the best public math pre-training dataset with 50B+ tokens!
HuggingFaceTB/finemath

Math remains challenging for LLMs and by training on FineMath we see considerable gains over other math datasets, especially on GSM8K and MATH.

We build the dataset by:
🛠️ carefully extracting math data from Common Crawl;
🔎 iteratively filtering and recalling high quality math pages using a classifier trained on synthetic annotations to identify math reasoning and deduction.

We conducted a series of ablations comparing the performance of Llama-3.2-3B-Base after continued pre-training on FineMath and observe notable gains compared to the baseline model and other public math datasets.

We hope this helps advance the performance of LLMs on math and reasoning! 🚀
We’re also releasing all the ablation models as well as the evaluation code.

HuggingFaceTB/finemath-6763fb8f71b6439b653482c2