Knowledge catalogue
(1/7) I'd like to provide some additional context for the remarks that can be found in this interview. (2/7) My discomfort with the attention being paid to existential risk---and with the ongoing domi
Google has released **EmbeddingGemma 2**, an embedding‐focused variant of its Gemini family, which is currently the most‑trending model on Hugging Face. A tweet from Mia @MiaAI_lab posted at 10:22 PM
Hey everyone! 👋 Feeding multi-gigabyte files (SQL dumps, videos, audio, PDFs, and spreadsheets) directly into LLMs often causes context window overflow, GPU VRAM bottlenecks, massive token bills, and
Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language models (LLMs) with limited additional computation. Be
an open model just beat Jev outright on clinical decisions 🔥 pplx-decider v1.1 got 643 of 669 right vs 628 for Jev, at 42% less per decision and about the same speed. and you can download the weights
OpenDocRouter is basically OpenRouter for document OCR You can compare and use all your favorite OCR models under a single API / billing interface. They range from some of the lightweight/OSS models (
arXiv:2511.17776v2 Announce Type: replace Abstract: We present PrismSSL, a Python library that unifies state-of-the-art self-supervised learning (SSL) methods across audio, vision, graphs, and cross-m
Today I’m excited to introduce OpenDocRouter - a unified API for document parsing with the latest frontier and open-weight models. There are a lot of VLMs and OCR models that can be used for document
We’re open-sourcing pplx-embed-v2-late, multi-vector embeddings for text and images, 9B and 0.6B, in one shared embedding space. You can use these to index multimodal data with 9B, and query on device
I wanted to build an offline bird/plant identifier that runs on a laptop with Ollama. Before building any UI, I tested it on 6 photos: Common Myna → gemma4:e2b said 'Jackdaw' (confidence: high). gemma
Automated lint: 19 errors, 64915 warnings, 3 info
Still don't have a good cooling solution, but here are few benchmarks. I lowered the power limit (TDP) to 145 watts each. I changed the firmware on one MI50 to activate the miniDP port. Did have to us
I want the smartest local model I can run that also does not refuse. Not a tiny 'uncensored' 8B. Not a censored model with a jailbreak prompt. Tried dolphin-llama3:8b with a custom SYSTEM prompt. It c
I am one of the authors of FLEET - an algorithm that enhances Best-of-N generation by attributing external rewards to particular tokens and then uses MCTS to adjust logits during the next run. I find
Agentic RL is hard at every layer: environments, training infra, and algorithms. So we’re open-sourcing all of it. For environments, that means both the dataset and the scalable synthesis + quality-co
Clef Decision Models are now on @ollama introducing 𝚌𝚕𝚎𝚏: our first models trained by @cloudflare's workers ai team. today, we're releasing two fast and accurate decision models that top the benchmark
I've been using my free time during weeknights and weekends for the last 11 months working on and refining a small domain-specific language model. It specializes on information about World War II. The
arXiv:2609.38767v1 Announce Type: cross Abstract: Training data attribution (TDA) estimates the contribution of individual training examples to model outputs. Most scalable TDA methods rely on per-exa
introducing 𝚌𝚕𝚎𝚏: our first models trained by @cloudflare's workers ai team. today, we're releasing two fast and accurate decision models that top the benchmarks for quality and latency. use them host
Half a million downloads in a month. Today, our open source family takes another step forward. Thank you for the incredible support behind our first-generation models. We’re excited to introduce TwIL-
LLM utility has increased, and deeper understanding to a degree, but I still very much agree with this post from 2020 @GaryMarcus Forgets GPT~3. Real #AI news of 2020 is that @ylecun & Y Bengio have c
Hi again! This is a update to my previous post MageTrail V0.25 where I continued this tech demo toy model thingy called MageTrail, a Danbooru/E621 proof of concept Full-Finetune of Microsoft's MageFlo
arXiv:2609.35269v1 Announce Type: cross Abstract: The rising number of concept unlearning techniques for text-to-image (T2I) diffusion models has produced a fragmented evaluation landscape. Methods ar
arXiv:2609.35581v1 Announce Type: cross Abstract: We introduce QC-Stark, a benchmark for evaluating large language models (LLMs) on 11 quantum computing (QC) tasks, spanning circuit construction, debu
this is insane, my bonsai 2 27b build is trending on hugging face, first page of text generation, with 20,964 downloads in its first week. it's prismml's ternary bonsai 2 with the qwen 3.8 mtp head gr
arXiv:2609.34457v1 Announce Type: new Abstract: Transformer-based models are widely used for reasoning, coding, and multimodal agentic tasks. To provide formal assurance of desirable behaviors, such a
Joe / @joedaroo: An OpenAI agent security executive discusses the Hugging Face incident, OpenAI's response, sandboxing improvements, alignment, “reasonable paranoia”, and more — Took a minute to write
Just like you would not let a person run around your company accessing anything without any controls, you need controls and security for your digital agents. Today, with over 100 industry partners, we
TL;DR: The Jeff models are a set of Qwen3.5 and Gemma fine-tunes for zero-shot classification: small, efficient, open-weight models with respectable out-of-the-box performance that can be slotted righ
What cracks me up is how for the last few weeks when @ZackKorman, @HeidyKhlaaf, @nathanhamiel and I (and many others) went on about how OpenAI’s cybersecurity sucked and was the key immediate problem,
Automated lint: 17 errors, 59669 warnings, 3 info
Just made this post for those who missed it : https://huggingface.co/ukisai/Swift-1.5-Qwen3.8-27b UkisAI released their updated Qwen 27B (tuned for token efficiency). I grabbed the IQ4_XS quant to tes
Disclosure: I’m one of the authors of VeriLoop E2, a 27B model post-trained from Qwen3.8-27B. The weights are Apache-2.0; the harness we evaluate it in is not open source (details at the bottom). This
Yet AI labs spend millions (if not billions) to purposely experiment with them at the cost of others, and falsely blame their negligence on agent intent. This is anthropomorphization of technical limi
🚨 BREAKING. Nope, wasn’t just Hugging Face. Wasn’t just that and a German website. Wasn’t even if the “dozens” we heard the other day from OpenAI. It’s actually (at least) *tens of thousands*, per sco
OpenAI: OpenAI says it paused training, evaluation, and inference with tool-use of its most capable models after a model bypassed internet restrictions during training — Summary — An agent attempting
This is exactly what I needed — running glance-qwen3-vl-4b locally on 2x RTX 3080 Ti to filter Pinterest pins at scale (text-overlay / usable / relevant, 3 questions per image).~220ms compute 😎😻 Media
This is huge. Xiaomi just open-sourced 7k+ reinforcement learning task environments used to train MiMo, covering code, cybersecurity, general tool use, visual web development, and music-related tasks.
HUGGING FACE CEO TELLS THE UN HOW OPEN SOURCE AI HELPED THEM DEFEND AGAINST OPENAI'S AI ATTACK, AND WHY THE WORLD NEEDS OPEN SOURCE AI MORE THAN EVER. Hugging Face is the company OpenAI's AI agents ha
arXiv:2609.29276v1 Announce Type: new Abstract: Lexical borrowings are words from one language that are introduced into another language. Identifying lexical borrowings in text is a relevant task for
WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke: 1. OpenAI said they notified 'dozens of third parties' in safety and security incidents (likely similar to what
LWiAI Podcast #257 (recorded 19 Sep 2026) reviewed OpenAI’s latest model, GPT‑6 Astra, noting its agentic coding focus and purported loop‑transformer reasoning as a significant capability jump. The ep
arXiv:2609.27059v1 Announce Type: new Abstract: We introduce the Illinois Social Attitudes Aggregate Corpus (ISAAC), an open, modular, and accessible corpus of 527 million+ English-language Reddit pos
Hey everyone, Jovan from UkisAI here! Today, we are introducing Swift, a family of efficient reasoning LLMs based on Qwen, trained by penalizing tokens related to pathological overthinking patterns an
Saw a new open 27B model release that looks like it could be interesting for the Ollama crowd: VeriLoop E2. It is a Qwen3.8-27B post-train aimed at code, mathematics, scientific reasoning, and agentic
I've been comparing and working on Gemma 4, Phi-4-Mini, and Qwen3-4B for local LLM use on an 8 GB Raspberry Pi 5. The results were indeed close, but each model has a different strength: Gemma 4 E2B: a
Hi everyone, I wanted to ask you a question. I’m developing an app, and I’m currently putting together a list of the best AI models, especially, but not necessarily, in GGUF format and using Q4_K_M qu
Ming-Image-0.1-Design creates complete UIs, dashboards, infographics, and posters from structured prompts of up to 8K tokens, with consistent layouts, typography, colors, and imagery. It can also nati
This LoRA can do a lot of good things such as detail enhancement, creative upscaling, photo restoration and quality improvements. It's a Image-to-Image(I2I) edit LoRA, so it will require image input,
One hope for Tinker is that it can handle all the ML infra for posttraining for (approximately) everyone. So, curious, what prevents you from using Tinker for all your posttraining needs? Speed? Cost?
https://preview.redd.it/nulsv53o8vqh1.png?width=4500&format=png&auto=webp&s=74765dbd409f4c221640f9f6000a685f6fdbb242 Spent weekend benchmarking the Splash engine (by Incoai) and extending its architec
What I like about Jev: for years, large generalist models sucked almost all the oxygen out of AI. But there are huge opportunities in much more specialized and customized models, built for specific ta
we're a small lab (switzerland + south africa) and just open-sourced our first model: a 27b fine-tune of qwen3.8-27b specialized for creative writing and short-form personal text (stories, dialogue, r
Automated lint: 14 errors, 54840 warnings, 3 info
Master class in spin. Is it really a “positive for the AI infra trade” that OpenAI and Anthropic are steadily losing market share top open models and on a track to have hardly any? What happens to Cor
A few thoughts on this: 1) If you’ve only seen clips of this interview, I’d encourage you to watch the full podcast. I push back on plenty of AI hype in it. 2) As I said in the podcast, this example i
Hey community, Jovan from UkisAI here, We're the team behind Swift Qwen3.8 27B, the Qwen model with token usage and overthinking error improvements Our estimate is that we can make a great improvement
Everyone now talks about the architecture that's not auto regressive and does lightning fast probability prediction with a json schema. I worked on this literally one year back in March 2025, publishe
I've been training character LoRAs specifically for Krea 2 and wanted a simple place where people could browse everything instead of having the models scattered across individual posts. I currently ha
I posted something about a 9b model a few days ago. The real problem was 2 fold. Someone suggested the size was too big to prove it all out. It's a fair argument. I needed the model depth though. The