Knowledge catalogue
Microsoft MAI models. Coming soon to Fireworks. Intelligence you control. End-to-end lineage you can prove. Fine-tune MAI reasoning models for your enterprise tasks. Your data. Your custom models. You
Pretty refreshing for foundation model training in 2026. I wonder how much of the capability gap between their model and other frontier models is because of this constraint. from: https://microsoft.ai
arXiv:2408.10441v3 Announce Type: replace Abstract: For many low-resource languages, the only available language models are large multilingual models trained on many languages simultaneously. Despite
arXiv:2602.03639v2 Announce Type: replace Abstract: Sampling-based controllers, such as Model Predictive Path Integral (MPPI) methods, offer substantial flexibility but often suffer from high variance
arXiv:2602.11137v2 Announce Type: replace-cross Abstract: Large language models are typically trained in two broad phases: pretraining to produce a base model, followed by further training to improve
arXiv:2605.29548v1 Announce Type: new Abstract: Larger models learn tasks smaller models do not. What drives this phenomenon? We develop a simple phenomenological argument that power-law scaling alrea
Noah McLaughlin created a 7-billion parameter language model and published it on Hugging Face, drawing a humorous parallel to how AI practitioners evaluate models using model cards by suggesting human
As Chinese AI models go closed source, and all U.S. frontier models are closed, there is a massive opportunity for a western open source AI Lab A future lack of competitive open source AI models is a
arXiv:2605.25418v1 Announce Type: cross Abstract: Generating 3D models from face sketches is an active topic of research in Computer Graphics due to its potential to tremendously facilitate the modeli
arXiv:2602.10134v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are pretrained on corpora containing trillions of tokens and, therefore, inevitably memorize sensitive informatio
We are releasing Carbon: a crazy fast DNA model Carbon is 275x faster than the next best model. So fast you can process the whole human genome on a single GPU in <2 days. Here are the tricks we used:
arXiv:2605.13740v1 Announce Type: new Abstract: Whether navigating a building, operating a robot, or playing a game, an agent that acts effectively in an environment must first learn an internal model
arXiv:2605.11854v1 Announce Type: new Abstract: Diffusion Language Models (DLMs) have recently emerged as a promising alternative to autoregressive language models, offering stronger global awareness
arXiv:2605.10447v1 Announce Type: cross Abstract: Agent-based models (ABMs) are increasingly used in macroeconomics, but their analysis still often relies on ad hoc Monte Carlo campaigns with heteroge
the deeper point: model choice and harness choice aren't independent variables. benchmarking a model without specifying the harness is like benchmarking a chip without specifying the compiler. the num
arXiv:2605.00994v1 Announce Type: new Abstract: Finetuning can significantly modify the behavior of large language models, including introducing harmful or unsafe behaviors. To study these risks, rese
arXiv:2605.02684v1 Announce Type: new Abstract: Spectral-based machine learning models have been increasingly deployed in chemometrics and spectroscopy, where predictive accuracy is as important as ex
open weight models are having a moment @masondrxy is making deepagents cli a coding harness that works amazing with open weight models what is missing? what should he add? try it out! deepagents-cli i
The year of the harness 🤝 Harnesses are our most direct layer to turn a model into a great product experience for users by shaping model behavior towards a goal This matters today because models have
arXiv:2604.22018v1 Announce Type: cross Abstract: Several brain foundation models (FM) have recently been proposed to predict brain disorders by modelling dynamic functional connectivity (FC). While t
Spot on 🎯 We don't need frontier intelligence to automate searches and sending emails - We don't need trillion parameter models to be able to summarize articles or technical documents - We don't need
arXiv:2503.17239v3 Announce Type: replace-cross Abstract: Fine-tuning large language models (LLMs) is a common practice to adapt generalist models to specialized domains. However, recent studies show
arXiv:2510.18457v3 Announce Type: replace Abstract: The performance of Latent Diffusion Models (LDMs) is critically dependent on the quality of their visual tokenizers. While recent works have explore
Vincent Chow / South China Morning Post: Tencent releases Hy3-preview, its first AI model developed under former OpenAI researcher Yao Shunyu; the model features 295B parameters, down from HY2's 400B
arXiv:2510.26782v3 Announce Type: replace-cross Abstract: A world model is an internal model that simulates how the world evolves. Given past observations and actions, it predicts the future physical
arXiv:2510.07248v3 Announce Type: replace Abstract: Small language models (SLMs) enable scalable tool-augmented multi-agent systems where multiple SLMs handle subtasks orchestrated by a powerful coord
I tested @huggingface ml-intern, given the prompt 'Fine-tune a Segment Anything Model (SAM) on a useful medical dataset. Train the model, and provide a comprehensive tutorial in a Jupyter Notebook fil
arXiv:2604.07484v1 Announce Type: cross Abstract: Generative reward models (GRMs) have emerged as a promising approach for aligning Large Language Models (LLMs) with human preferences by offering grea
arXiv:2604.06483v1 Announce Type: cross Abstract: Large language models that require multiple GPU cards to host are usually the most capable models. It is necessary to understand and steer these model
arXiv:2503.13551v5 Announce Type: replace Abstract: Recent studies show that Large Language Models (LLMs) achieve strong reasoning capabilities through supervised fine-tuning or reinforcement learning
Anthropic investigated the internal mechanisms of its latest unreleased model, Claude Mythos Preview, and what they found is 100% worth a read. Key things I pulled from Anthropic researchers' threads:
.@steipete has been an amazing supporter of open models and local models. This is even before the rise of OpenClaw. 🦞 Ollama is here to love. ❤️ Let's make the world a better place together. Some folk
arXiv:2608.10107v1 Announce Type: new Abstract: Emerging World-Action Models (WAMs) have demonstrated promising performance in autonomous driving by jointly modeling future driving scene evolution and
arXiv:2608.10837v1 Announce Type: cross Abstract: The strong performance of foundation models for tabular tasks comes at substantial inference costs. Distilling models into task-specific architectures
NVIDIA Nemotron 3.5 Lighting is available on Ollama! It's a 30B model made for always-on agents. All local. Claude Code ollama launch claude --model nemotron-3.5-lightning Hermes Agent ollama launch h
A few months ago, I had the idea of making a LLM from scratch as a personal project (for learning and partly for improving my resume). Since I learned a lot from other posts on here over the past year
LFM2.5-2.6B is a new tiny model by LiquidAI, with benchmarks that put it head to head with much larger models. I've run llama-perplexity on many model GGUF quants, crossed with many KV cache quants, t
For a very long time most high-performing AI models were end-to-end neural models; vector input -> vector output, with only ultra-thin symbolic preprocessing and postprocessing layers (e.g. label deco
New York Times: Developers in Africa are increasingly choosing Chinese open-source AI models over US models, saying they are downloadable, easier to customize, and much cheaper — Developers built Sunf
arXiv:2607.27797v1 Announce Type: new Abstract: Predictive process monitoring (PPM) leverages event logs to forecast the future of running process instances, for instance, predicting the next activity
I was surprised to see that Deepseek V4 Flash is extremely smart and small enough to fit in setup that can be built with < $50,000. Expensive, but not a datacenter. So I wanted to see the trend over t
Introducing the Parse Gateway There's been an explosion of interest in model routing - you don't always need the best model for every task. That is especially true for document parsing 📄🔀: - Some page
arXiv:2607.23976v1 Announce Type: cross Abstract: Appending a two-word confirmation tag to a decision question -- 'Is X the better choice?' versus 'X is the better choice, right?' -- changes whether a
arXiv:2607.22535v1 Announce Type: cross Abstract: Action-conditioned video world models predict future observations from an initial observation and an action signal. In robotics, actions influence fut
arXiv:2607.22034v1 Announce Type: cross Abstract: Vision-language models (VLMs) are increasingly deployed on consumer hardware where input images are degraded by compression, camera shake, and poor li
There is a lot of buzz around running 100B parameter models on cheap local hardware using ternary (1.58-bit) quantization like Microsoft's BitNet architecture. The theoretical hardware shortcuts are i
Yes, open-source / open-weight models are important for a healthy AI ecosystem. That's how we can verify things, check claims, and keep up outside the closed labs. Plus, it gives us the freedom to run
Model quantization has become essential for efficient large language model deployment, yet existing approaches involve clear trade-offs: methods such as GPTQ and AWQ achieve practical compression but
Interesting finding on frontier models. It turns out that frontier models can write proofs but stumble on faithfully copying a long block of text. This has huge implications. It sounds trivial, which
I think OpenRouter is not a good measure of actual model usage in a world of agentic tools (not that I doubt that Chinese open weights model usage is up, but this could also look like a graph of usage
arXiv:2607.07729v1 Announce Type: cross Abstract: As foundation models grow in scale and diversity, coordinating multiple models into cooperative reasoning systems offers a path toward safer, more rel
Cursor announced that Grok 4.5 and Composer represent distinct model weight classes, with Composer 2.5 continuing to be available while new models in that size category will be released. This indicate
arXiv:2607.02642v1 Announce Type: new Abstract: Evaluating embodied robot foundation models remains a critical bottleneck; unlike large language models efficiently assessed via digital benchmarks, rob
arXiv:2607.04360v1 Announce Type: cross Abstract: Conditional generative models have emerged as powerful tools for sampling from target conditional distributions, driving substantial advances across a
arXiv:2607.03347v1 Announce Type: new Abstract: We consider the Multiscale Single-Index Model (MSIM), first introduced in ite{oymak2021learning}, as a stylized model for hierarchical learning with sca
Databricks introduces Vibe Data Modeling, a new approach to data modeling designed specifically for lakehouse architectures that aims to simplify and improve how organizations structure and manage dat
Dean Takahashi / GamesBeat: Tripo AI, which is developing 3D foundation models and world models for use in games, raised 150M, a month after it said it raised 200M — Tripo AI, a global artificial inte
Alex Stamos / @alexstamos: [Thread] The US' now-lifted export controls on Anthropic models created uncertainty, are an own goal, and will push service providers towards Chinese AI models — A lot to un
Watch Devin Fusion work: The sidekick model handles fetching information and skimming sources. The frontier model makes a plan for implementation, then delegates well-scoped sub-tasks to the sidekick
arXiv:2601.07965v2 Announce Type: replace Abstract: When a model knows when it does not know, many possibilities emerge. The first question is how to enable a model to recognize that it does not know.