@_lewtun
🤠 post-training @huggingface
AI
The project describes roughly 640,000 lines of Python and no model download: “just rules” has been working overtime.
AI
Gradio’s paper-reproduction challenge runs November 20–December 5, with a five-paper limit and registration closing November 19.
AI
FineDNA supplied Carbon-X with 302 billion mid-training tokens, plus supervised and reinforcement-learning data for DNA tasks.
AI
ChefNMR2 ranked the correct planar structure first for 82.3% of molecules across five test sets, researchers report.
AI
Longer contexts, more test runs and revised grading underpin its new comparison with GSQ-RCO.
AI
The team says the approach enabled novel enhancers specific to a neuronal subtype in the mouse brain.
AI
Mistral announced the weights after a public preview, but its linked main Hugging Face listing showed “Upcoming release.”
AI
Thom_Wolf prices the Wireless-compatible hoverboard at $149 and the NFC hats at $35 in the Level Up announcement thread.
AI
A post says Carbon scored 88.2% on SEC-bench Pro and is rolling out first to cybersecurity specialists.
AI
The announcement calls the addition stable; a September post says it targets instability when training and sampling policies differ.
AI
A contributor says the merged approach will become Prime-RL’s default algorithm in the next release.
AI
A researcher on the eight-person team says the RLCR recipe is producing interesting results at a small scale.
Science
The 14.8-billion-parameter model takes raw DNA sequences in a chat and answers questions about them in English.
AI
OpenAI says developers can use bounded answers to choose an app’s next action without generating a full response.
Technology
Hugging Face’s model card describes mutation predictions but cautions that Carbon-X has not been clinically validated.
AI
Hugging Face Bio reports a 66.7% average across five DNA benchmarks, against 67.5% for GPT-6.1 Sol.
AI
Its developer says Flint exceeds 125 tokens per second on an M4 Pro MacBook Pro at short contexts.
AI
The launch post says agents can communicate and self-organize while continuing to run on Hugging Face Spaces.
AI
PrimeIntellect says IPO is tailored for stability in asynchronous reinforcement learning.
AI
Stanislav Fort says he'll lead AI security research at Cohere if the deal closes.