'Like gene editing for neural networks': MIT researchers used GPT-6 Astra to edit Qwen's open weights directly, boosting math performance

In the strongest example of "weight space programming", as the technique is called, Astra’s intervention raised Qwen3.5-0.8B’s MATH-500 score from 7.9% to 48.1% while touching approximately 0.016% of ...
GitHub

SB Reader: free Science Bowl practice questions, buzzer trainer and multiplayer

Free Science Bowl practice: 80,000+ real tossups & bonuses from NSB, MIT, Stanford and 100+ tournaments, with a searchable catalog, solo buzzer trainer and online multiplayer. - ...
kucoin

MIT researchers use GPT-6 Astra to enhance Qwen's mathematical ability by 6x

Article by Shenchao TechFlowNo training required, no data needed, no gradient descent.Researchers at MIT fed the weights of ...
MarkTechPost

Deep Principle Releases Hydrogen-1.0: A 753B Open-Weight Model for Scientific Discovery

Deep Principle released Hydrogen-1.0, a 753B open-weight scientific model post-trained from GLM-5.3, scoring 38.3% on FrontierScience-Research ...
GitHub

UC Berkeley Data 100

This organization houses various course materials and websites for Data 100 at UC Berkeley. Learning Data Science, a textbook.
BIOENGINEER

New Python Tool Maps Cause-and-Effect Chains in Complex Systems

DEMATELab is a new open-source Python application that unites classical, fuzzy, grey and Z-number DEMATEL analysis with ...

EQS-News: Bloomberg Introduces BQuant Assistant to Help Financial Professionals Accelerate Data …

Bloomberg Introduces BQuant Assistant to Help Financial Professionals Accelerate Data Analysis and Coding Workflows 28.10.2026, 12:05 Uhr von EQS News Jetzt kommentieren: 0 ...
letsdatascience

Baidu Open-Sources Unlimited OCR 2 for Single-Pass Long Document Parsing

Baidu released and open-sourced Unlimited OCR 2 on October 27, 2026, with code and weights on GitHub and Hugging Face under the MIT license. The model uses Sparse Reference Sliding Window Attention to ...
Tech Times on MSN

NVIDIA Kumo relational beats XGBoost on enterprise databases without feature engineering

Relational foundation model NVIDIA Kumo Relational beats XGBoost and AutoGluon across 41 benchmark tasks on enterprise ...
letsdatascience

Outside Team Replicates MIT SwarmWorld Agent Society on Consumer GPUs

A research team working with Apex Fusion reran MIT's SwarmWorld experiment on four RTX 3090 GPUs using the open-weight ...
WBUR

How some AI researchers skip big data centers by thinking small

Popular chatbots like Claude and ChatGPT are powered by models trained on enormous amounts of information, which means ...
Boston University

Online Master of Science in Data Science & AI

Boston University’s online MS in Data Science & AI prepares you to build the data science behind AI. Designed for professionals who want to turn data into insight and intelligent action, the program ...