My colleagues – the talented scientists and engineers behind the Apertus models @swiss-ai – continue to resolve issues and respond to questions about the 1.5 release, at the same time as we continue work on the roadmap to 2.0. Today we celebrate our next milestone: the new Technical Report has months of evaluation and analysis in one heap of a preprint 📚 See our new post for a summary, and please 🤗 upvote! https://huggingface.co/blog/swiss-ai/apertus-v15-release-notes
1 reply
·
reactedtoBanaxi-Tech'spost with 👍about 1 month ago
Most things you do on HuggingFace, BananaMindBot can do. Fast
Mention @BananaMindBot on a model, dataset, Space discussion, paper, blog comment, or top-level post and it'll reply there.
It's powered by North Code Mini (Qwen3.8 27B, with GPT OSS 120B as fallback).
A few things it can do:
Search for models and datasets Look up users and orgs and see what they've published Read model cards, configs, dataset files, blog posts, and org profiles Answer questions about what it finds Write and run its own code in a locked-down sandbox when it needs to verify something Check things like a model's real parameter count from the safetensors headers instead of just repeating the model card Remember something for later if you explicitly ask it to Forward a message to @Banaxi-Tech Post a daily roundup of developments in the small-language-model space
It won't execute code you give it. It can read and review that code, but anything it runs is code it wrote itself.
It also can't access private data or credentials.
Mention it somewhere.
It's going to also find this post!
(Some parts inspired by CompactBot and @CompactAI Follow them please)
OpenRouter Leaderboard — every model, every provider, one comparable table. Price, precision, uptime, measured latency and language quality on the same axes.
Building it turned up three things.
We graded 330 models on Korean and two axes collapsed.
Honorifics — only 8.5% earn an A Knowledge of Korean institutions — 9.4% Every other axis sits above 31% Fluency hides it. A model can write clean, natural Korean and still attach an honorific to a coffee cup. Fluent and wrong at the same time is worse than obviously broken, because nobody catches it in review.
A 2023 model beats the 2026 flagships. gpt-3.5-turbo-16k scores a perfect 3.00. Korean cannot be inferred from release date, parameter count or English benchmarks — it has to be measured, per model.
Quality, value and speed are three different models. Across five axes, the same model almost never takes two columns.
425 models, latency measured on 329 on a paid API, Korean graded on 330. Three languages, three currencies, daily refresh, open API, no key.
Wohoo 🥳 I have finished my 2025 GPU workstation build and I am very excited to train new awesome open source models on it.
I built my last GPU workstation 5 years ago featuring an AMD Ryzen 5900X, 64GB of G.SKILL Trident Z RGB on an ASRock X570 Taichi cooled by an Alphacool Eisbär 420. GPU was a Zotac RTX 3090 AMP Extreme. Unfortunately, I was never satisfied with the case - some Fractal Define 7, as it is definitely too small, airflow is not optimal as I had to open the front door all the time and it also arrived with a partly damaged side panel.
For my new build, I've used the following components: an outstanding new AMD Ryzen 9950X3D with 64GB of Corsair Dominator Titanium (what a name). As a huge Noctua fan - warm greetings to my Austrian neighbors - I am using the brand new Noctua NH-D15 G2 on an ASRock X870E Taichi in an amazing Lian Li LANCOOL III chassis. One joke that only NVIDIA Blackwell users will understand: you definitely need a tempered glass panel to check if your GPU cables/connectors start melting 😂 And the best is yet to come: I returned my previously bought Zotac RTX 5090 Solid to the eBay seller (because of... missing ROPs, only NVIDIA Blackwell users will again understand) and bought a Zotac 5090 AMP Extreme INFINITY (yes, the long name indicates that this is the flagship model from Zotac) from a more trustworthy source (NBB in Germany).
I am so happy to start training and fine-tuning new open source models - stay tuned!!!
🚀 We are delighted to announce MamayLM, a new state-of-the-art efficient Ukrainian LLM!
📈 MamayLM surpasses similar-sized models in both English and Ukrainian, while matching or overtaking up to 10x larger models.
📊 MamayLM is a 9B model that can run on a single GPU, enabling cost-efficient AI autonomy and adoption across sectors in Ukraine such as education, legal, healthcare, public services and others (e.g., by specializing it to particular use cases). MalayLM is also attractive for organizations wishing to preserve data privacy as it s efficiency allows it to run on a local machine.
🧠 MamayLM is trained on high-quality Ukrainian data and understands Ukrainian language, culture, and history. It is built on top of Google’s Gemma 2 9B model, but uses a number of new advances stemming from INSAIT’s experience in creating BgGPT, a Bulgarian LLM we released last year, now adopted nationwide and profiled several times by Google as a worldwide success case.
🤝 MamayLM is developed in a collaboration between researchers at INSAIT and ETH Zürich and is trained entirely via donations to INSAIT for AI compute resources.
📥 MamayLM is now freely available to download on INSAIT’s HuggingFace in both full and quantized versions. We also publicly release all Ukrainian benchmarks we evaluated on.
📝 Further, we release blog posts in both English and Ukrainian, sharing our approach to creating MamayLM, hoping to drive further improvements by the community.
🌎 The release of LLMs for various languages is part of INSAIT’s mission in ensuring countries can achieve AI autonomy in a cost-efficient, controlled, safe and predictable manner.
Translating benchmarks is a painful process, requiring a lot of manual inspection and adjustments. You start from setting up the whole pipeline and adapting to every format type, including task specifics. There already exist some massive benchmarks, but they still have some simple (and sometimes silly) bugs, which can hurt the evaluations :( We present a novel automated translation framework to help with that!
Eastern and Southern European languages introduce richer linguistic structures compared to English and for benchmarks which heavily rely on grammatical coherence machine translation presents a risk of harming evaluations. We discover potential answer leakage or misleading through grammatical structure of the questions. Some benchmarks are also just outdated and need to be retranslated with newer and better models.
We present a framework with novel test-time scaling methods which allow to control time and cost investments, while at the same time mitigate the need for human-in-the-loop verification. While working on Ukrainian-focused MamayLM models, we had to translate 10+ benchmarks in a short span of time. Finding human evaluators is costly and time-consuming, same goes for using professional translators. With our pipeline we were able to do it in 3 days🏎️
We hope our findings will help enable stronger multilingual evaluations and developments. We release all produced benchmarks on Hugging Face together with the source code and Arxiv paper 🤗
if you like it give the demo a little star and send a shoutout to : @MaxLSB@jddqd and @GAD-cell for absolutely obliterating the pareto frontier of the french language understanding .
Here is one of the equations that make up the worlds first Artificial General Intelligence. Remember when building Artificial Intelligence or anything on a device it all starts out binary. Everything starts out with data flow physics and mathmatics
"sonnet 5 drops tomorrow and i've heard from three separate sources inside anthropic that the benchmarks they're sitting on would mass-retire every model released in 2025. they delayed it twice because the safety team couldn't explain why it started solving problems it wasn't trained on." (https://x.com/iruletheworldmo/status/2019237039904878902)
Update: Making My AI Recruiting Assistant More Deterministic, Auditable, and Bias-Aware
Hi everyone — I wanted to share a progress update on my AI recruiting assistant and some recent changes focused on reliability and transparency. The goal of this project is to build a decision-support tool for recruiters that doesn’t just “sound confident,” but can actually explain why it produces a given recommendation.
We built an agent to surface and implement high-potential ideas for your repo, asynchronously generating containers, tests, and PRs so you can evaluate what works and double down on it.
FastVLMs by Apple are the talk of the week for edge device VLMs and also for consumer-grade VLMs on the Hub. They have some impressive demos available on the Hub for live captioning and inference tasks. Meanwhile, I’m still exploring one of the coolest edge-device multimodal releases—Liquid AI’s LFM2-VL (450M and 1.6B). I’ve also made a live camera video inference demo, which is capable of running on Colab’s free-tier T4 GPU.
🌲🍄 LLM Forest Orchestra: Turning Hidden States into Music
Hello everyone! I'm excited to introduce a new Space I've been developing called LLM Forest Orchestra. This project converts the hidden states and attention patterns of transformer models into layered MIDI compositions. The concept draws inspiration from mushrooms and mycelial networks in forests. Fungi create underground connections linking plants and trees, establishing what some call a "wood-wide web" where signals and nutrients travel. Researchers have discovered that these exchanges form patterns resembling rhythms and pulses. When translated appropriately, these patterns can become music.
Transformers operate through remarkably similar principles: tokens share signals via hidden states and attention heads. This Space transforms those invisible information flows into notes, chords, and rhythms, treating the model as a digital forest orchestra.
🎛 Features
* Two compute modes: - Full model operates on a Hugging Face model (defaulting to unsloth/Qwen3-14B-Base). - Mock latents provides a CPU-friendly option that simulates tensors for immediate experimentation. * Musical controls: You can adjust scale selection, tempo grid, velocity range, instrument/role presets, and seed randomization. * Output: The system generates .mid files compatible with DAWs and remixing workflows.
🌌 Why?
Neural networks already resemble unusual musical instruments: signals flow through them, patterns emerge organically, and careful observation reveals hidden melodies. This is analogous to the forest's secret orchestra of mushrooms and trees.
👉 Try it
Try the Space here: Locutusque/LLM-Forest-Orchestra. I'm excited to hear the sounds you can generate. Please share your created MIDIs or remixes in the comments. Let's explore how this hidden forest of transformers can sound together. 🌳🎶
Published a stable version of Ukrainian Text-to-Speech library on GitHub and PyPI.
Features:
- Multi-speaker model: 2 female (Tetiana, Lada) + 1 male (Mykyta) voices; - Fine-grained control over speech parameters, including duration, fundamental frequency (F0), and energy; - High-fidelity speech generation using the RAD-TTS++ acoustic model; - Fast vocoding using Vocos; - Synthesizes long sentences effectively; - Supports a sampling rate of 44.1 kHz; - Tested on Linux environments and Windows/WSL; - Python API (requires Python 3.9 or later); - CUDA-enabled for GPU acceleration.
🌟 Progress in the German FineWeb edu reproduction 🌟
We're delighted to share the launch of our new Data Quality Classification Model, designed specifically for evaluating educational content in German. This tool uses advanced machine learning techniques to assess texts across all educational levels, from primary school to university.
🔍 Inspired by Huggingface's fine web edu dataset, we've worked hard to refine data classification methods ensuring educators and learners access top-quality resources. We're excited about the future as we continue improving our models and expanding our datasets.
🙏 A huge thank you to David and Daryoush from Vago Solutions; Björn and Jan from Ellamind / DiscoResearch for their expert insights throughout this project. Your support has been crucial. This project was made possible by the support of PrimeLine AI.
