43 129 279

Gabriel Martín Blázquez

gabrielmbmb

https://gabrielmb.com

AI & ML interests

ML Engineer

Recent Activity

reacted to anton-l's post with 🚀 4 days ago

Introducing 📐𝐅𝐢𝐧𝐞𝐌𝐚𝐭𝐡: the best public math pre-training dataset with 50B+ tokens! https://huggingface.co./datasets/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. https://huggingface.co./collections/HuggingFaceTB/finemath-6763fb8f71b6439b653482c2

updated a dataset 5 days ago

gabrielmbmb/gsm8k-reasoning-paths-combined

upvoted a paper 5 days ago

Qwen2.5 Technical Report

View all activity

Articles

How we leveraged distilabel to create an Argilla 2.0 Chatbot

Jul 16

• 32

Organizations

gabrielmbmb's activity

reacted to anton-l's post with 🚀 4 days ago

Post

1964

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

updated a dataset 5 days ago

gabrielmbmb/gsm8k-reasoning-paths-combined

Viewer • Updated 5 days ago • 100 • 20

upvoted a paper 5 days ago

Qwen2.5 Technical Report

Paper • 2412.15115 • Published 6 days ago • 327

liked a Space 6 days ago

Running

103

🏃

Jupyter Agent

liked a model 6 days ago

answerdotai/ModernBERT-base

Fill-Mask • Updated 6 days ago • 27.6k • 437

upvoted a paper 6 days ago

Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference

Paper • 2412.13663 • Published 7 days ago • 103

reacted to burtenshaw's post with 🤗❤️ 6 days ago

Post

2540

People are flexing their end of year stats, so I made this app to show hub stats in a tidy design!

Thanks @Ameeeee and @jfcalvo for the feature from Argilla!
burtenshaw/recap