Michael Pieler

MicPie's activity

• All
• Models
• Datasets
• Spaces
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

upvoted a collection about 2 months ago

ChemPile

Collection

The ChemPile is a dataset with over 77 billion curated multimodal tokens about chemistry. For more information, visit https://chempile.lamalab.org/. • 8 items • Updated 20 days ago • 16

upvoted a paper 2 months ago

Thought Anchors: Which LLM Reasoning Steps Matter?

Paper • 2506.19143 • Published Jun 23 • 13

upvoted an article 3 months ago

view article

Article

Towards Open Evolutionary Agents

By and 1 other •

Aug 4

• 18

upvoted an article 5 months ago

view article

Article

System Prompt Learning: Teaching LLMs to Learn Problem-Solving Strategies from Experience

By •

Jun 2

• 20

liked a dataset 8 months ago

Josephgflowers/Par-Four-Fineweb-Edu-Fortified-Chemistry-Physics-Astronomy-Math-Reason

Viewer • Updated Nov 16, 2024 • 988k • 67 • 5

liked a Space about 1 year ago

Running

295

295

GPU Poor LLM Arena

🏆

Compact LLM Battle Arena: Frugal AI Face-Off!

upvoted a collection about 1 year ago

Open LLM Leaderboard best models ❤️‍🔥

Collection

A daily uploaded list of models with best evaluations on the LLM leaderboard: • 65 items • Updated Mar 20 • 648

liked 2 Spaces over 1 year ago

Runtime error

38

38

StableLM 2 12B Chat

📉

Generate text responses to user queries

Runtime error

101

101

Stable Code Instruct 3b

📉

Chat with an AI assistant for detailed responses

reacted to m-ric's post with 👀 over 1 year ago

view post

Post

2271

𝗧𝗵𝗲 𝗵𝘂𝗴𝗲 𝗰𝗼𝘀𝘁 𝗼𝗳 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗼𝗻 𝗳𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗟𝗟𝗠𝘀 💸

Google DeepMind recently released a great paper that shows optimal hyperparameters to train across different regimes: Scaling Exponents Across Parameterizations and Optimizers, with data from 10,000 training runs.

One engineer decided to quantify the price of such a large-scale experiment.

😬 And the bill is hefty: ~13M USD

This exact number is to take with a grain of salt because many approximations were necessary to get the final result.

⛔️ But still this ballpark means that for this sole experiment, the price is way over what most startups or research labs could afford.

This means that open-sourcing research is more important than ever, to put everyone in the ecosystem on a roughly equal footing. Don't let OpenAI run first, they'll keep everything for themselves!

Read the full post that quantifies the paper's cost 👉 https://152334h.github.io/blog/scaling-exponents/

• 

• 
• 1 reply

·

👀

5

5

🔥

1

1

+

Reply

New activity in HuggingFaceFW/fineweb-edu-llama3-annotations over 1 year ago

100 word justifications from Llama3 annotations

2

#3 opened over 1 year ago by

MicPie

liked a dataset over 1 year ago

HuggingFaceFW/fineweb-edu-llama3-annotations

Viewer • Updated Jun 3, 2024 • 467k • 286 • 45

liked a Space over 1 year ago

Running

1.17k

1.17k

FineWeb: decanting the web for the finest text data at scale

🍷

Generate high-quality text data for LLMs using FineWeb

upvoted a collection over 1 year ago

📀 Dataset comparison models

Collection

1.8B models trained on 350BT to compare different pretraining datasets • 8 items • Updated Jun 12, 2024 • 40

liked a model over 1 year ago

sfairXC/FsfairX-LLaMA3-RM-v0.1

Text Classification • 8B • Updated Oct 14, 2024 • 1.36k • 60

authored 4 papers over 1 year ago

Stable LM 2 1.6B Technical Report

Paper • 2402.17834 • Published Feb 27, 2024 • 3

Are large language models superhuman chemists?

Paper • 2404.01475 • Published Apr 1, 2024 • 19

GPT-NeoX-20B: An Open-Source Autoregressive Language Model

Paper • 2204.06745 • Published Apr 14, 2022 • 1

Few-shot Adaptation Works with UnpredicTable Data

Paper • 2208.01009 • Published Aug 1, 2022

New activity in JeanKaddour/minipile about 2 years ago

Domain and provenance annotation

9

#1 opened about 2 years ago by

haukur