Asking because about once every six months I put out a call for examples of fine tuning LLMs working in practice and, aside from the obvious examples like Cursor, I don't often get many great answers
OPEN SOURCE
-
State Space Models in Gallery Updates
By
–
I have models with state space layers in the gallery (e.g., Nemotron 3). Anything since April I missed though?
-
Local AI Community Discusses Open Weights and Inference
By
–
The first Local AI Get-Together was a massive success This pic is missing quite a few people who left before we hit the 4-hour mark, but thank you to everyone who stopped by Local AI is very real, very alive, and apparently willing to talk GPUs, open weights, inference
-
Architectural design of open-source AI libraries
By
–
真没见过哪个开源 Repo，能把 18 个 use case 塞进去还没乱套。
— 艾略特 (@elliotchen100) 10 mai 2026
架构方法、Benchmark、Memory 模式，一库打尽。 https://t.co/9z2G8VxrSvI've never seen an open-source repo that can cram 18 use cases in without turning into a mess. Architecture methods, benchmarks, memory patterns—all covered by one library.
-
Gary Marcus on AGI and scaling limits
By
–
So many people misremember (or never read) what I said in in 2022 in “Deep learning is hitting a wall”, which was neither about revenue or AI’s potential upper limits. Rather, it was an argument that the pure of scaling LLMs would not get us to AGI, and that we would need to
-
MiniCPM-o 4.5 for omni-modal interactions
By
–
MiniCPM-o 4.5 Towards Real-Time Full-Duplex Omni-Modal Interaction paper: https://
huggingface.co/papers/2604.27
393
… -
DeepSeek v4 Flash Impresses Locally
By
–
OPEN SOURCE AI is killing it. DeepSeek v4 Flash is a quasi-frontier model with a massive 1M context window. It can LOCALLY on a 128GB Mac using specialized 2-bit quantization. Asked my OpenClaw Engineering Agent Ross about the model and he's impressed.
-
Open-Sourced Sparse Training Code for H100 GPUs
By
–
For those interested in the implementation details, we’ve open-sourced the reference code for this paper. The repository includes our sparse training code and the custom CUDA kernels designed for H100 GPUs leveraging the TwELL packing format. GitHub:
-
Reflecting on the progress of open-source AI
By
–
August through October will be for the history books when it comes to Opensource AI The past couple of months might feel huge but I believe we haven’t seen the good stuff of 2026 yet
-
METR’s Graph: 50% Success, Not 100
By
–
Hot take on METR’s new graph that so many people are flipping about today. • Claude Code is a real advance; Mythos probably builds on some of what is learned there. But… • If you read the graph carefully, it is about achieving *50%* success. Not 100 or 99 or even 90. The