It comes in two sizes: 1.2B and 350M I'm a big fan of the 350M model that can be used on GPUs to do big data operations. It's an absolute banger that shows how powerful fine-tuning can be
MACHINE LEARNING
-
Tiny Task-Specific AI Models for Edge Devices
By
–
We're releasing a collection of tiny task-specific models Want to do data extraction, translation, RAG, tool use, or math on a Raspberry Pi? We got you covered! Here are a few examples ↓
-
ShinkaEvolve: Cost-Efficient AlphaEvolve Implementation and Comparison
By
–
ShinkaEvolve is our implementation of AlphaEvolve (we try to make it more sample efficient to save costs). Here is a comparison between Darwin Gödel Machine and AlphaEvolve from @richardcsuwandi
’s blog post discussing both methods: https://
richardcsuwandi.github.io/blog/2025/dgm/ -
ShinkaEvolve: Open-Source Framework for Scientific Program Evolution
By
–
Proud to release ShinkaEvolve, our open-source framework that evolves programs for scientific discovery with very good sample-efficiency! 🐙
— hardmaru (@hardmaru) 25 septembre 2025
Paper: https://t.co/05rjMiSxOL
Blog: https://t.co/yjuvV0xpif
Project: https://t.co/DpSEScrAws https://t.co/2pmiPesV4KProud to release ShinkaEvolve, our open-source framework that evolves programs for scientific discovery with very good sample-efficiency! Paper: https://
arxiv.org/abs/2509.19349
Blog: https://
sakana.ai/shinka-evolve/
Project: https://
github.com/SakanaAI/Shink
aEvolve
… -
Meta FAIR Introduces Code World Model for Advanced Code Generation Research
By
–
New from Meta FAIR: Code World Model (CWM), a 32B-parameter research model designed to explore how world models can transform code generation and reasoning about code.
— AI at Meta (@AIatMeta) 24 septembre 2025
We believe in advancing research in world modeling and are sharing CWM under a research license to help empower… pic.twitter.com/kwykTJu1GFNew from Meta FAIR: Code World Model (CWM), a 32B-parameter research model designed to explore how world models can transform code generation and reasoning about code. We believe in advancing research in world modeling and are sharing CWM under a research license to help empower
-
AI Uncovers Hidden Failures in Predictive Maintenance
By
–
Standard maintenance tools catch the obvious failures. But real downtime hides in subtle, slow-developing failures. Discover how @AI4ProductionOutcomes uncovers what others miss. Read more: https://
buff.ly/lr9ESOF #sponsored #infinite_iiot #PredictiveMaintenance #PrescriptiveAI -
Siemens leads AI-powered manufacturing data alliance initiative
By
–
Big news from #EMO2025: @Siemens joins forces with machine builders in a data alliance to reshape manufacturing with AI.
Details https://
buff.ly/CDo7796 [Sponsored] #SIE_EMO #SmartManufacturing @TRUMPFinc @HELLER_Group_EN @siemensindustry -
Databricks Data Science Agent Automates Analytics Workflows
By
–
Our new Data Science Agent transforms Databricks Assistant into an autonomous partner for data science and analytics.
— Databricks (@databricks) 24 septembre 2025
Turn hours of work into minutes while staying in control
Integrated into Notebooks and the SQL Editor, this agent explores data, trains models, fixes errors,… pic.twitter.com/3JquEmLaoJOur new Data Science Agent transforms Databricks Assistant into an autonomous partner for data science and analytics. Turn hours of work into minutes while staying in control Integrated into Notebooks and the SQL Editor, this agent explores data, trains models, fixes errors,
-
Adding Datasets to Hugging Face Platform
By
–
Nice! Any chance for it to be added to http://
hf.co/datasets? -
Scaling Language Models with Infinite Compute via Ensemble Distillation
By
–
As data and compute scales differently, what do we do when we have infinite compute? To answer this, this paper shows you can still scale LMs by cranking up regularization, training many small clones on the same corpus, then ensemble → distill, netting 5x data-efficiency