The Diff & The Merge: my new substack over at oxykodit.substack.com/ where I write about the day-to-day life of being an open-source maintainer in the Python/data/AI domain. Posts so far include: – Monthly January 2026 update (Typer reference docs, uv, nanochat progress) – How to get the most out of your Open-Source contributions? – Deploying an agent swarm to improve LLM training code The last post being a fun Saturday activity toying with a swarm variant of @karpathy's autoresearch repo published last week. Will Agent OxyKodit manage to lower the val_bpb significantly, and set a new master record? Will there actually be a Monthly February update as well? Subscribe to find out 😅
MACHINE LEARNING
-
Continual Learning Experience Skills Multimodal Agents
By
–
Continual Learning from Experience and Skills for Multimodal Agents
-
Global AI App Layer Race Continues with Industry Leaders
By
–
The global race for the AI app layer is still on #AI #AIio #AIInnovation #ML #DataScience #Futureofwork @fabiomoioli @pascal_bornet @alliekmiller @mattshumer_ @OfficialLoganK @jeremyphoward @GaryMarcus ow.ly/nUKM30sTyFt
→ View original post on X — @terence_mills, 2026-03-14 21:00 UTC
-
Dense Task Experts in Pretrained Neural Networks Thickets
By
–
RL is no longer needed? "Neural Thickets: Diverse Task Experts Are Dense Around Pretrained Weights" This paper argues that large pretrained models don’t sit at a single optimal set of weights but inside a dense “thicket” of nearby task-specific experts. So once pretraining is
-
Neural Cellular Automata Pre-Training for Language Models
By
–
Pre-Pre-Training on Neural Cellular Automata? This paper argues that LLMs may not need human text to learn core thinking machinery. So by first training on synthetic neural cellular automata, the model learns transferable rule-and-pattern inference that speeds up and improves
-
One person cured cancer with $3,000 and ChatGPT subscription.
By
–
$3,000 and a ChatGPT subscription cured cancer That’s what it cost one person to do what used to require a full oncology research lab, institutional funding, and years of grant applications. The DNA sequencing was $3,000. The protein structure prediction (AlphaFold) is free.
-
Two Fast LLM Inference Optimization Techniques Explained
By
–
Two different tricks for fast LLM inference https://
buff.ly/FYkJWvm
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
FlowRVS: Language-Guided Video Object Segmentation
By
–
What if AI could precisely segment any object in a video, guided only by your natural language description? A joint effort by SGIT AI Lab, UC San Diego, HKUST, UTokyo, Cambridge, Zhejiang University of Technology, and Baidu presents FlowRVS. Instead of finding then
-
AI successfully animates game characters, facing initial load challenges
By
–
Nice but too weird too basic not enough effort put by you I think
-
AlphaFold Recognized as Greatest AI Achievement for Future Generations
By
–
We will look back on AlphaFold as one of the greatest things to come from AI. Will keep giving for generations to come.