Hugging Face TB Research

Enterprise
community
Activity Feed

AI & ML interests

Exploring smol models and high quality web and synthetic datasets, generated by LLMs (TB is for Textbook, as inspired by the "Textbooks are all your need" paper)

Recent Activity

HuggingFaceTB's activity

merve 
posted an update about 18 hours ago
view post
Post
662
Everything that happened this week in open AI, a recap 🤠 merve/jan-17-releases-678a673a9de4a4675f215bf5

👀 Multimodal
- MiniCPM-o 2.6 is a new sota any-to-any model by OpenBMB
(vision, speech and text!)
- VideoChat-Flash-Qwen2.5-2B is new video multimodal models by OpenGVLab that come in sizes 2B & 7B in resolutions 224 & 448
- ByteDance released larger SA2VA that comes in 26B parameters
- Dataset: VRC-Bench is a new diverse benchmark for multimodal LLM reasoning performance

💬 LLMs
- MiniMax-Text-01 is a new huge language model (456B passive 45.9B active params) by MiniMaxAI with context length of 4M tokens 🤯
- Dataset: Sky-T1-data-17k is a diverse dataset used to train Sky-T1-32B
- kyutai released Helium-1-Preview-2B is a new small multilingual LM
- Wayfarer-12B is a new LLM able to write D&D 🧙🏻‍♂️
- ReaderLM-v2 is a new HTML parsing model by Jina AI

- Dria released, Dria-Agent-a-3B, new agentic coding model (Pythonic function calling) based on Qwen2.5 Coder
- Unsloth released Phi-4, faster and memory efficient Llama 3.3

🖼️ Vision
- MatchAnything is a new foundation model for matching
- FitDit is a high-fidelity VTON model based on DiT architecture

🗣️ Audio
- OuteTTS-0.3-1B is a new multilingual text-to-speech model with voice cloning and emotion control capabilities

📖 Retrieval
- lightblue released a new reranker based on Qwen2.5 LB-reranker-0.5B-v1.0 that can handle 95+ languages
- cde-small-v2 is a new sota small retrieval model by
@jxm
merve 
posted an update 1 day ago
Xenova 
posted an update 1 day ago
view post
Post
1612
Introducing Kokoro.js, a new JavaScript library for running Kokoro TTS, an 82 million parameter text-to-speech model, 100% locally in the browser w/ WASM. Powered by 🤗 Transformers.js. WebGPU support coming soon!
👉 npm i kokoro-js 👈

Try it out yourself: webml-community/kokoro-web
Link to models/samples: onnx-community/Kokoro-82M-ONNX

You can get started in just a few lines of code!
import { KokoroTTS } from "kokoro-js";

const tts = await KokoroTTS.from_pretrained(
  "onnx-community/Kokoro-82M-ONNX",
  { dtype: "q8" }, // fp32, fp16, q8, q4, q4f16
);

const text = "Life is like a box of chocolates. You never know what you're gonna get.";
const audio = await tts.generate(text,
  { voice: "af_sky" }, // See `tts.list_voices()`
);
audio.save("audio.wav");

Huge kudos to the Kokoro TTS community, especially taylorchu for the ONNX exports and Hexgrad for the amazing project! None of this would be possible without you all! 🤗

The model is also extremely resilient to quantization. The smallest variant is only 86 MB in size (down from the original 326 MB), with no noticeable difference in audio quality! 🤯
  • 2 replies
·
MoritzLaurer 
posted an update 3 days ago
view post
Post
1763
Microsoft's rStar-Math paper claims that 🤏 ~7B models can match the math skills of o1 using clever train- and test-time techniques. You can now download their prompt templates from Hugging Face !

📏 The paper introduces rStar-Math, which claims to rival OpenAI o1's math reasoning capabilities by integrating Monte Carlo Tree Search (MCTS) with step-by-step verified reasoning trajectories.
🤖 A Process Preference Model (PPM) enables fine-grained evaluation of intermediate steps, improving training data quality.
🧪 The system underwent four rounds of self-evolution, progressively refining both the policy and reward models to tackle Olympiad-level math problems—without GPT-4-based data distillation.
💾 While we wait for the release of code and datasets, you can already download the prompts they used from the HF Hub!

Details and links here 👇
Prompt-templates docs: https://moritzlaurer.github.io/prompt_templates/
Templates on the hub: MoritzLaurer/rstar-math-prompts
Prompt-templates collection: MoritzLaurer/prompt-templates-6776aa0b0b8a923957920bb4
Paper: https://arxiv.org/pdf/2501.04519
ngxson 
posted an update 4 days ago
view post
Post
1571
Check out my collection of pre-made GGUF LoRA adapters!

This allow you to use both normal + abliterated version of popular models like llama, qwen, etc, without having to double to amount of VRAM usage.

ngxson/gguf_lora_collection
·
davanstrien 
posted an update 5 days ago
view post
Post
2901
Introducing scandi-fine-web-cleaner davanstrien/scandi-fine-web-cleaner, the first model trained on FineWeb-C community annotations!

FineWeb2 is a massive multilingual dataset for pre-training language models. Like any web-scale dataset, it contains low-quality content. How can we improve it?

Over the past months, an amazing community of 400+ annotators has been labelling content quality (using Argilla) across 23 languages through the FineWeb-C initiative.

Today, I'm happy to share the first classifier trained on this data.

🔍 What we've built:

- A lightweight classifier that efficiently removes low-quality content
- 90%+ precision demonstrated on Danish & Swedish
- Can process the 43M+ documents in Danish FineWeb2 with minimal compute

🌍 Why this matters: The approach can be reproduced for any of the 23 languages in FineWeb-C ( data-is-better-together/fineweb-c). We can improve training data quality at scale without massive compute resources by starting with community annotations and training small, efficient classifiers.

Want to build a classifier for your language? Check out the full blog post with code examples and implementation details: https://danielvanstrien.xyz/posts/2025/FineWeb-c/scandinavian-content-filtering-fineweb.html
  • 1 reply
·
merve 
posted an update 5 days ago
view post
Post
3778
there's a new multimodal retrieval model in town 🤠
LlamaIndex released vdr-2b-multi-v1
> uses 70% less image tokens, yet outperforming other dse-qwen2 based models
> 3x faster inference with less VRAM 💨
> shrinkable with matryoshka 🪆
> can do cross-lingual retrieval!
Collection: llamaindex/visual-document-retrieval-678151d19d2758f78ce910e1 (with models and datasets)
Demo: llamaindex/multimodal_vdr_demo
Learn more from their blog post here https://huggingface.co/blog/vdr-2b-multilingual 📖
MoritzLaurer 
posted an update 7 days ago
view post
Post
2918
FACTS is a great paper from @GoogleDeepMind on measuring the factuality of LLM outputs. You can now download their prompt templates from @huggingface to improve LLM-based fact-checking yourself!

📏 The paper introduces the FACTS Grounding benchmark for evaluating the factuality of LLM outputs.

🤖 Fact-checking is automated by an ensemble of LLM judges that verify if a response is fully grounded in a factual reference document.

🧪 The authors tested different prompt templates on held-out data to ensure their generalization.

📚 It's highly educational to read these templates to learn how frontier labs design prompts and understand their limitations.

💾 You can now download and reuse these prompt templates via the prompt-templates library!

🔄 The library simplifies sharing prompt templates on the HF hub or locally via standardized YAML files. Let’s make LLM work more transparent and reproducible by sharing more templates like this!

Links 👇
- prompt-templates docs: https://moritzlaurer.github.io/prompt_templates/
- all templates on the HF Hub: MoritzLaurer/facts-grounding-prompts
- FACTS paper: https://storage.googleapis.com/deepmind-media/FACTS/FACTS_grounding_paper.pdf
merve 
posted an update 8 days ago
view post
Post
3539
What a beginning to this year in open ML 🤠
Let's unwrap! merve/jan-10-releases-677fe34177759de0edfc9714

Multimodal 🖼️
> ByteDance released SA2VA: a family of vision LMs that can take image, video, text and visual prompts
> moondream2 is out with new capabilities like outputting structured data and gaze detection!
> Dataset: Alibaba DAMO lab released multimodal textbook — 22k hours worth of samples from instruction videos 🤯
> Dataset: SciCap captioning on scientific documents benchmark dataset is released along with the challenge!

LLMs 💬
> Microsoft released Phi-4, sota open-source 14B language model 🔥
> Dolphin is back with Dolphin 3.0 Llama 3.1 8B 🐬🐬
> Prime-RL released Eurus-2-7B-PRIME a new language model trained using PRIME alignment
> SmallThinker-3B is a new small reasoning LM based on Owen2.5-3B-Instruct 💭
> Dataset: QWQ-LONGCOT-500K is the dataset used to train SmallThinker, generated using QwQ-32B-preview 📕
> Dataset: @cfahlgren1 released React Code Instructions: a dataset of code instruction-code pairs 📕
> Dataset: Qwen team is on the roll, they just released CodeElo, a dataset of code preferences 👩🏻‍💻

Embeddings 🔖
> @MoritzLaurer released zero-shot version of ModernBERT large 👏
> KaLM is a new family of performant multilingual embedding models with MIT license built using Qwen2-0.5B

Image/Video Generation ⏯️
> NVIDIA released Cosmos, a new family of diffusion/autoregressive World Foundation Models generating worlds from images, videos and texts 🔥
> Adobe released TransPixar: a new text-to-video model that can generate assets with transparent backgrounds (a first!)
> Dataset: fal released cosmos-openvid-1m Cosmos-tokenized OpenVid-1M with samples from OpenVid-1M

Others
> Prior Labs released TabPFNv2, the best tabular transformer is out for classification and regression
> Metagene-1 is a new RNA language model that can be used for pathogen detection, zero-shot embedding and genome understanding
davanstrien 
posted an update 8 days ago
view post
Post
2074
The data-is-better-together/fineweb-c dataset is growing!

This week a few more languages have got 1,000 annotations for the educational quality of data from HuggingFaceFW/fineweb-2.

Why should you care?

The quality of pre-training data can have a big impact on the performance of downstream language models trained on that data ( HuggingFaceFW/blogpost-fineweb-v1).

Being able to filter by educational quality is on way of improving the quality of the data you use for training an LLM. Very importantly this approach can also reduce the amount of data needed for pertaining.

Why not use an LLM?

LLMs can be used to annotate educational quality for a subset of data. This data can then be used to train a smaller encoder only model to label the full dataset. However, this may not work well for languages outside of english. This is where fineweb-c (community) comes in.

The community is annotating the educational quality of fineweb2 data. Currently 114 languages have some annotations. These annotations will enable a number of things:

- Evaluate whether an LLM can label the educational quality for texts in that language well
- Directly be used for training quality classifiers
- Help discover other rules and huerisitcs for refining fineweb2 further for different languages.

This week the following languages where done:

Swedish thanks to: @Lauler @AntonVic @ohallstrom @bjarlestam @menbom @Ekgren @apsod

Ukrainian thanks to: @hannayukhymenko @robinhad @realPivo @RabotiahovDmytro @reciprocate

Assamese thanks to: @moyoor97 @Arpanjyoti @nawaf-helmi123 @pahigogoi1 @aelhence @kishorekashyap

Want to learn more: https://huggingface.co/blog/davanstrien/fineweb2-community

Contribute yourself here: data-is-better-together/fineweb-c
  • 1 reply
·
cfahlgren1 
posted an update 8 days ago
view post
Post
1340
Wow, I just added Langfuse tracing to the Deepseek Artifacts app and it's really nice 🔥

It allows me to visualize and track more things along with the cfahlgren1/react-code-instructions dataset.

It was just added as a one click Docker Space template, so it's super easy to self host 💪
MoritzLaurer 
posted an update 9 days ago
view post
Post
1678
The TRL v0.13 release is 🔥! My highlight are the new process reward trainer to train models similar to o1 and tool call support:

🧠 Process reward trainer: Enables training of Process-supervised Reward Models (PRMs), which reward the quality of intermediate steps, promoting structured reasoning. Perfect for tasks like stepwise reasoning.

🔀 Model merging: A new callback leverages mergekit to merge models during training, improving performance by blending reference and policy models - optionally pushing merged models to the Hugging Face Hub.

🛠️ Tool call support: TRL preprocessing now supports tool integration, laying the groundwork for agent fine-tuning with examples like dynamic temperature fetching in prompts.

⚖️ Mixture of judges: The new AllTrueJudge combines decisions from multiple binary judges for more nuanced evaluation.

Read the release notes and other resources here 👇
Release: https://github.com/huggingface/trl/releases/tag/v0.13.0
Mergekit: https://github.com/arcee-ai/mergekit
Mixture of judges paper: The Perfect Blend: Redefining RLHF with Mixture of Judges (2409.20370)
merve 
posted an update 9 days ago
view post
Post
1746
ByteDance just dropped SA2VA: a new family of vision LMs combining Qwen2VL/InternVL and SAM2 with MIT license 💗 ByteDance/sa2va-model-zoo-677e3084d71b5f108d00e093

> The models are capable of tasks involving vision-language understanding and visual referrals (referring segmentation) both for images and videos ⏯️

> The models come in 1B, 4B and 8B and are based on InternVL2.5 for base architecture and Qwen2, Qwen2.5 and InternLM2 for language model part (depending on the checkpoint)

> The model is very interesting, it has different encoders for different modalities each (visual prompt, text prompt, image and video) then it concatenates these to feed into LLM 💬

the output segmentation tokens are passed to SAM2, to sort of match text (captions or semantic classes) to masks ⤵️

> Their annotation pipeline is also interesting, they seems to use two open large vision LMs to refine the annotations, and have different levels of descriptions to provide consistency.
  • 1 reply
·
ngxson 
posted an update 10 days ago
MoritzLaurer 
posted an update 11 days ago
view post
Post
2045
OpenAI is losing money on the $200/month subscription 🤯. It's crazy how expensive it is to run these largest LLMs:

- ChatGPT Pro costs $200/month ($2,400/year) and is still unprofitable for OpenAI due to higher-than-expected usage.
- OpenAI reportedly expected losses of about $5 billion on revenue of $3.7 billion last year, with ChatGPT alone once costing an estimated $700,000 per day to operate. 💸🔥
- They build strong models and do great research. Whether this business model will work in the long run is one of the biggest questions in the AI economy today.

Source with the numbers 👇
https://techcrunch.com/2025/01/05/openai-is-losing-money-on-its-pricey-chatgpt-pro-plan-ceo-sam-altman-says/
·
MoritzLaurer 
posted an update 12 days ago
view post
Post
2191
🚀 Releasing a new zeroshot-classifier based on ModernBERT! Some key takeaways:

- ⚡ Speed & efficiency: It's multiple times faster and uses significantly less memory than DeBERTav3. You can use larger batch sizes and enabling bf16 (instead of fp16) gave me a ~2x speed boost as well
- 📉 Performance tradeoff: It performs slightly worse than DeBERTav3 on average across my zeroshot classification task collection
- 🧠 Use cases: I recommend using it for scenarios requiring speed and a larger context window (8k).
- 💡 What’s next? I’m preparing a newer version trained on better + longer synthetic data to fully leverage the 8k context window and improve upon the training mix of my older zeroshot-v2.0 models. I also hope that there will be a multilingual variant in the future.

Great work by https://huggingface.co/answerdotai !

If you’re looking for a high-speed zeroshot classifier, give it a try!

📄 Resources below: 👇
Base model: MoritzLaurer/ModernBERT-base-zeroshot-v2.0
Large model: MoritzLaurer/ModernBERT-large-zeroshot-v2.0
Updated zeroshot collection: MoritzLaurer/zeroshot-classifiers-6548b4ff407bb19ff5c3ad6f
ModernBERT collection with paper: answerdotai/modernbert-67627ad707a4acbf33c41deb