calculatortamer

🇨🇳 Gitee Code Dataset - The Missing Piece of the Stack
nyuuzyou/gitee-code

Gitee is not included in the Software Heritage archive, meaning it is currently missing from datasets like The Stack. This release fills that massive gap, serving as the largest Chinese code dataset and one of the largest code corpuses overall.

- 819,472,785 files from 3,105,923 repositories
- 536 GB compressed Parquet storage
- 554 programming languages
- Extensive quality filtering: Removed vendor code, artifacts, and generated files
- Rich Chinese language understanding: High volume of Chinese comments and docs

Huge thanks to Hugging Face for the storage grant that made hosting this (and all my other datasets) possible!

I have also already dropped several other new code datasets and rolled out QoL improvements for older ones. I will be dropping posts on those throughout the week.

1 reply

upvoted an article 11 months ago

Article

Continuous batching from first principles

ror, ArthurZ, mcpotato

•

Nov 25, 2025

• 450

replied to m-ric's post 11 months ago

it's been a while i stopped everything
can i resume my activities?

reacted to m-ric's post with 🚀 about 1 year ago

Post

5142

STOP EVERYTHING NOW - we might finally have a radical architecture improvement over Transformers!!! 🚨

A lone scientist just proposed Tiny Recursive Model (TRM), and it is literally the most impressive model that I've seen this year.

➡️ Tiny Recursive Model is 7M parameters
➡️ On ARC-AGI, it beats flagship models like Gemini-2.5-pro

Consider how wild this is: Gemini-2.5-pro must be over 10,000x bigger
and had 1,000 as many authors 😂 (Alexia is alone on the paper)

What's this sorcery?
In short: it's a very tiny Transformers, but it loops over itself at two different frequencies, updating two latent variables: one for the proposed answer and one for the reasoning.

@AlexiaJM started from the paper Hierarchical Reasoning Model, published a few months ago, that already showed breakthrough improvement on AGI for its small size (27M)

Hierarchical Reasoning Model had introduced one main feature:
🔎 Deep supervision
In their model, one part (here one layer) would run at high frequency, and another would be lower frequency, running only every n steps.

They had used a recurrent architecture, where these layers would repeat many times ; but to make it work they had to do many approximations, including not fully backpropagating the loss through all layers.

Alexia studied what was useful and what wasn't, and cleaned the architecture as follows :
Why use a recurrent architecture, when you can just make it a loop?
➡️ She made the network recursive, looping over itself

Why use 2 latent variables ?
➡️ She provides a crystal clear explanation : the one that changes frequently is the reasoning, the one that changes at low frequency is the proposed answer.
➡️ She runs ablation studies to validate that 2 is indeed optimal.

This new setup is a much more elegant way to process reasoning than generating huge chains of tokens as all flagship models currently do.

This might be the breakthrough we've been awaiting for so long!