Some new updates to Papers with Code [P]
Hi folks, Niels here from the open-source team at Hugging Face. I continue working on a revival of paperswithcode.co as we're back to the "age of…
Communities 142 stories
GPU access in 2026 is still fragmented — is there a better market structure for compute? [P]
Anyone building at the model layer knows the procurement problem hasn't gone away. H100s are still allocated unevenly, spot instances get preempted at the worst times,…
[ECCV 2026] Paper Decision Appeals Discussion [D]
With the release of meta-reviews, ECCV sent out a google form for dissatisfied authors to submit an appeal for the following reasons: Policy errors, e.g., reviewers…
An Update on Matrix Recurrent Units, an Attention Alternative [R]
I recently revisited my matrix recurrent units algorithm (the MRU), a novel linear-time sequence architecture I created as an alternative to attention. I explain it in…
Data-centric debugging for teams training neural nets [P]
We just did a big revamp of WeightsLab and wanted to share it here. If you’ve ever spent hours debugging a training run only to discover…
Best current methods for finetuning whisper on domain specific vocabulary? [P]
Hey everyone, I’m wondering whether there are any newer or more effective methods for fine tuning whisper on domain specific speech. I’m working on a project…
EMA on LoRA ? [R]
Hi guys Does anyone know of papers where EMA on LoRA adapters has been used successfully? Im interested in cases where the EMA adapter acts as…
A slightly improved DVD-JEPA demo [P]
Hey! I came across this post, which I found quite neat as a minimal demonstration of JEPA. However, as the comments pointed out, there was some…
How transparent is DiffusionGemma (and why it matters)
Authors: Joshua Engels*, Callum McDougall*, Bilal Chughtai*, Janos Kramar, Senthoran Rajamanoharan, Cindy Wu, Arthur Conmy, Asic Q Chen, Jean Tarbouriech, Min Ma, Brendan O'Donoghue+, João Gabriel…
Hi Reddit, I posted my Build Your Own LLM workshop to Youtube teaching ML, LLM and math intuition [P]
Hi internet friends, I recorded a workshop about building your own LLM without any math / ML prerequisites. It covers everything from machine learning fundamentals, deep…
Would you let an ML PhD student graduate without a top-tier paper? [D]
Suppose you’re a PhD advisor in machine learning. Your student has been in the program for 4 years, has done solid work, and has a coherent…
Best Local Agents – Jun 2026
A megathread that is overdue! Let's discuss and debate on what the best local agents available today are Prologue First a note on terminology: While most…
GDM AI Control Roadmap
GDM has published an AI Control Roadmap! From the executive summary:We present the GDM AI Control Roadmap (v0.1) – our plan for implementing and adopting internal…
Predicting LLM Safety Before Release by Simulating Deployment
Paper linkBefore releasing a new model, labs need to understand not just what it can do, but how it is likely to behave in real-world use,…
Synthetic document finetuning for instilling positive traits
This is the fifth in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas. The fourth…
Why Do Naive SFT Filters For Safety Properties Fail?
This is the fourth in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas. The third…
SFT Drives Gemini’s Safety Properties
This is the third in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas. The second…
Building and evaluating model diffing agents
This is the second in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas. The first…
Sympathy for both sides of the egregious misalignment debate
On one side of this debate is Yudkowsky & Soares, who think that (if AI progress continues) we’re on a direct path to egregiously-misaligned, scheming, out-of-control,…
Models May Behave Worse When Eval Aware
This is the first in a series of research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas.TL;DRIt's often assumed that…
[D] Self-Promotion Thread
Please post your personal projects, startups, product placements, collaboration needs, blogs etc. Please mention the payment and pricing requirements for products and services. Please do not…
[D] Monthly Who’s Hiring and Who wants to be Hired?
For Job Postings please use this template Hiring: [Location], Salary:[], [Remote | Relocation], [Full Time | Contract | Part Time] and [Brief overview, what you're looking…