Anyone tried using the new (ish) Gemma diffusion model as a speculative model?
It seems that MTP is the gold standard for speed up but still suffers from having to choose between regressive and parallel drafters that come with…
Communities 6782 stories
Give Smart People the Tools to Do Smart Things
Article URL: https://superuserdone.com/posts/2026-07-03-give-smart-people-the-tools/ Comments URL: https://news.ycombinator.com/item?id=48775748 Points: 8 # Comments: 0
Mistral released Leanstral-1.5-119B-A6B
Leanstral 1.5, a free Apache-2.0 licensed model with 6B active parameters, delivers a major performance upgrade in formal verification, saturating miniF2F, solving 587/672 PutnamBench problems, and…
BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decoding
Speculative decoding accelerates inference by using a lightweight draft model to generate candidate tokens in parallel, and are then verified by the target model, enabling lossless…
Instead of banning AI, I made a classroom contract with my students
Article URL: https://www.science.org/content/article/instead-banning-ai-i-made-classroom-contract-my-students Comments URL: https://news.ycombinator.com/item?id=48775499 Points: 28 # Comments: 7
Show HN: TaskPeace – a task queue my AI coding agents pull work from over MCP
Article URL: https://taskpeace.com/ Comments URL: https://news.ycombinator.com/item?id=48775484 Points: 3 # Comments: 0
Anatomy of Persistent Memory's 3 Layers: Comparing ContextNest, Mem0 and Zep
Article URL: https://promptowl.ai/resources/persistent-memory-ai-agents/ Comments URL: https://news.ycombinator.com/item?id=48775483 Points: 4 # Comments: 0
ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving
In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models…
GLM 5.2 is really good!
I'm probably late to the party when it comes to reviewing GLM 5.2, but I've been using it recently and I'm impressed. My use case: I…
Micro-World – Action-controlled Interactive world model – AMD
In this work, we introduce Micro-World, an action-controlled interactive world model designed to generate high-quality, open-domain scenes. Built on top of the Wan2.1 family of models,…
Gemma Avatar: Talk to Gemma 4-31B face to face
This is a voice chat with Gemma 4 31B where you talk to a 3D avatar. It listens while you speak, answers with a voice and…
Pragmatic FDT, and predictors as game theory
Decision theory is back in fashion (defining fashion as "one good post on a good EA blog"). Bentham's Bulldog (BB) has published a case against FDT…
Fable #6: The Return of the King
The blip is over. We have Fable back. Utah teapot: happy fable/mythos easter Wednesday, to those who celebrate Here is the official letter restoring Fable, great…
Hackers shoveled snow for company, were rewarded with network admin access
Article URL: https://www.theregister.com/security/2026/07/02/hackers-shoveled-snow-for-company-were-rewarded-with-network-admin-access/5265240 Comments URL: https://news.ycombinator.com/item?id=48774704 Points: 9 # Comments: 0
Valve open source the Steam Machine e-ink screen so you can make your own
Article URL: https://www.gamingonlinux.com/2026/07/valve-open-source-the-steam-machine-e-ink-screen-so-you-can-make-your-own/ Comments URL: https://news.ycombinator.com/item?id=48774518 Points: 50 # Comments: 6
PostgreSQL and the OOM Killer: Why You Must Use Strict Memory Overcommit
Article URL: https://www.ubicloud.com/blog/postgresql-and-the-oom-killer-why-we-use-strict-memory-overcommit Comments URL: https://news.ycombinator.com/item?id=48774509 Points: 20 # Comments: 2
The Fall and Rise of Screwworm
Article URL: https://www.construction-physics.com/p/the-fall-and-rise-of-screwworm Comments URL: https://news.ycombinator.com/item?id=48774492 Points: 3 # Comments: 0
Zuckerberg 'Admits' Meta's Layoffs Were Ineffective
Article URL: https://eshumarneedi.com/2026/07/03/zuckerberg-admits-metas-layoffs-were.html Comments URL: https://news.ycombinator.com/item?id=48774454 Points: 54 # Comments: 24
Gemini Code Assist will be shut down on July 17
Article URL: https://docs.cloud.google.com/gemini/docs/code-review/review-repo-code Comments URL: https://news.ycombinator.com/item?id=48774429 Points: 8 # Comments: 0
Please stop the AI confidence theater
Article URL: https://www.elenaverna.com/p/please-stop-the-ai-confidence-theater Comments URL: https://news.ycombinator.com/item?id=48774414 Points: 67 # Comments: 33
My DeepSeek V4 Pro at home got faster again
You may remember my earlier posts about DeepSeek V4 Pro at home. Today I checked the performance in my llama.cpp branch that contains various fixes and…
Program-as-Weights: A Programming Paradigm for Fuzzy Functions
Article URL: https://arxiv.org/abs/2607.02512 Comments URL: https://news.ycombinator.com/item?id=48774368 Points: 7 # Comments: 0
Anyone looked into "Thermo-NN" optimizing model architecture via thermodynamic/Landauer cost instead of pure FLOPs?
Came across a project framing architecture search around thermodynamic information cost (Landauer's principle) rather than the usual FLOPs/params tradeoff, the pitch is causal derivation before implementation,…
AI Data Centers Use More Water Than Most Tech Giants Report
Article URL: https://www.wsj.com/tech/ai/ai-data-centers-water-use-901e2902 Comments URL: https://news.ycombinator.com/item?id=48774246 Points: 15 # Comments: 5