davanstrien (Daniel van Strien)

reacted to cbensimon's post with 🔥 19 days ago

Post

5714

🚀 ZeroGPU medium size is now available as a power-user feature

Nothing too fancy for now—ZeroGPU Spaces still default to large (70GB VRAM)—but this paves the way for:
- 💰 size-based quotas / pricing (medium will offer significantly more usage than large)
- 🦣 the upcoming xlarge size (141GB VRAM)

You can as of now control GPU size via a Space variable. Accepted values:
- auto (future default)
- medium
- large (current default)

The auto mode checks total CUDA tensor size during startup:
- More than 30GB → large
- Otherwise → medium

3 replies

·

reacted to their post with 🔥 about 1 month ago

Post

2203

Came across a very nice submission from @marcodsn for the reasoning datasets competition (https://huggingface.co/blog/bespokelabs/reasoning-datasets-competition).

The dataset distils reasoning chains from arXiv research papers in biology and economics. Some nice features of the dataset:

- Extracts both the logical structure AND researcher intuition from academic papers
- Adopts the persona of researchers "before experiments" to capture exploratory thinking
- Provides multi-short and single-long reasoning formats with token budgets - Shows 7.2% improvement on MMLU-Pro Economics when fine-tuning a 3B model

It's created using the Curator framework with plans to scale across more scientific domains and incorporate multi-modal reasoning with charts and mathematics.

I personally am very excited about datasets like this, which involve creativity in their creation and don't just rely on $$$ to produce a big dataset with little novelty.

Dataset can be found here: marcodsn/academic-chains (give it a like!)

posted an update about 1 month ago

Post

2203

Came across a very nice submission from @marcodsn for the reasoning datasets competition (https://huggingface.co/blog/bespokelabs/reasoning-datasets-competition).

The dataset distils reasoning chains from arXiv research papers in biology and economics. Some nice features of the dataset:

- Extracts both the logical structure AND researcher intuition from academic papers
- Adopts the persona of researchers "before experiments" to capture exploratory thinking
- Provides multi-short and single-long reasoning formats with token budgets - Shows 7.2% improvement on MMLU-Pro Economics when fine-tuning a 3B model

It's created using the Curator framework with plans to scale across more scientific domains and incorporate multi-modal reasoning with charts and mathematics.

I personally am very excited about datasets like this, which involve creativity in their creation and don't just rely on $$$ to produce a big dataset with little novelty.

Dataset can be found here: marcodsn/academic-chains (give it a like!)

reacted to jasoncorkill's post with 🔥 about 2 months ago

Post

3064

🔥 Yesterday was a fire day!
We dropped two brand-new datasets capturing Human Preferences for text-to-video and text-to-image generations powered by our own crowdsourcing tool!

Whether you're working on model evaluation, alignment, or fine-tuning, this is for you.

1. Text-to-Video Dataset (Pika 2.2 model):
Rapidata/text-2-video-human-preferences-pika2.2

2. Text-to-Image Dataset (Reve-AI Halfmoon):
Rapidata/Reve-AI-Halfmoon_t2i_human_preference

Let’s train AI on AI-generated content with humans in the loop.
Let’s make generative models that actually get us.

reacted to ajibawa-2023's post with 🔥 about 2 months ago

Post

4107

Hi All, I recently released two Audio datasets which are generated using my earlier released dataset: ajibawa-2023/Children-Stories-Collection

First Audio Dataset:https://huggingface.co/datasets/ajibawa-2023/Audio-Children-Stories-Collection-Large has 5600++ stories in .mp3 format.

Second Audio Dataset:https://huggingface.co/datasets/ajibawa-2023/Audio-Children-Stories-Collection has 600 stories in .mp3 format.

3 replies

·

reacted to jasoncorkill's post with 🚀🔥 about 2 months ago

Post

3282

🚀 We tried something new!

We just published a dataset using a new (for us) preference modality: direct ranking based on aesthetic preference. We ranked a couple of thousand images from most to least preferred, all sampled from the Open Image Preferences v1 dataset by the amazing @data-is-better-together team.

📊 Check it out here:
Rapidata/2k-ranked-images-open-image-preferences-v1

We're really curious to hear your thoughts!
Is this kind of ranking interesting or useful to you? Let us know! 💬

If it is, please consider leaving a ❤️ and if we hit 30 ❤️s, we’ll go ahead and rank the full 17k image dataset!

6 replies

·

replied to jasoncorkill's post about 2 months ago

This is very cool! I was always curious about doing something like this! Could be quite cool to train a "aesthic preference model" on this kind of dataset. Could be quite cool to try and use as a reward model for image gen training...

cc @sayakpaul @multimodalart @linoyts @davidberenstein1957 who might also find this data interesting :)

reacted to jasoncorkill's post with ❤️ about 2 months ago

Post

3282

🚀 We tried something new!

We just published a dataset using a new (for us) preference modality: direct ranking based on aesthetic preference. We ranked a couple of thousand images from most to least preferred, all sampled from the Open Image Preferences v1 dataset by the amazing @data-is-better-together team.

📊 Check it out here:
Rapidata/2k-ranked-images-open-image-preferences-v1

We're really curious to hear your thoughts!
Is this kind of ranking interesting or useful to you? Let us know! 💬

If it is, please consider leaving a ❤️ and if we hit 30 ❤️s, we’ll go ahead and rank the full 17k image dataset!

6 replies

·

posted an update about 2 months ago

Post

1688

I've created a v1 dataset ( davanstrien/reasoning-required) and model ( davanstrien/ModernBERT-based-Reasoning-Required) to help curate "wild text" data for generating reasoning examples beyond the usual code/math/science domains.

- I developed a "Reasoning Required" dataset with a 0-4 scoring system for reasoning complexity
- I used educational content from HuggingFaceFW/fineweb-edu, adding annotations for domains, reasoning types, and example questions

My approach enables a more efficient workflow: filter text with small models first, then use LLMs only on high-value content.

This significantly reduces computation costs while expanding reasoning dataset domain coverage.

posted an update 3 months ago

Post

2943

📊 Introducing "Hugging Face Dataset Spotlight" 📊

I'm excited to share the first episode of our AI-generated podcast series focusing on nice datasets from the Hugging Face Hub!

This first episode explores mathematical reasoning datasets:

- SynthLabsAI/Big-Math-RL-Verified: Over 250,000 rigorously verified problems spanning multiple difficulty levels and mathematical domains
- open-r1/OpenR1-Math-220k: 220,000 math problems with multiple reasoning traces, verified for accuracy using Math Verify and Llama-3.3-70B models.
- facebook/natural_reasoning: 1.1 million general reasoning questions carefully deduplicated and decontaminated from existing benchmarks, showing superior scaling effects when training models like Llama3.1-8B-Instruct.

Plus a bonus segment on bespokelabs/bespoke-manim!

https://www.youtube.com/watch?v=-TgmRq45tW4

reacted to stefan-it's post with 🔥 3 months ago

Post

3216

After running some 3DMark and FurMark benchmarks on Windows to make sure that my new 5090 is not causing melting cables [1] and some nice shots with a thermal camera (I don't think that's too much), running some fine-tuning experiments with my favorite Flair & Transformers libraries are very easy to perform.

Important steps:

Good idea is to start with a fresh Ubuntu 24.04 installation with latest CUDA 12.8 and the open NVIDIA driver - follow more advices from [2]:

sudo apt -y install cuda-toolkit-12-8 nvidia-open

I tried update from an existing Ubuntu installation with an older CUDA and driver version and it resulted in a non-startable system.

If you are using PyTorch 2.6 with built CUDA 12.6 it will result in:

NVIDIA Graphics Device with CUDA capability sm_120 is not compatible with the current PyTorch installation.
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_70 sm_75 sm_80 sm_86 sm_90.

But no worries! For PyTorch you need just to use a nightly 2.7 version that was built with CUDA 12.8. This can easily done via:

pip install --pre torch --index-url https://download.pytorch.org/whl/nightly/cu128

After that the latest Flair version can be installed and fine-tuning will work!

References:

[1]: https://www.reddit.com/r/nvidia/comments/1inpox7/rtx_50_series_12vhpwr_megathread/
[2]: https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=24.04&target_type=deb_network

1 reply

·

posted an update 3 months ago

Post

3692

Quick POC: Turn a Hugging Face dataset card into a short podcast introducing the dataset using all open models.

I think I'm the only weirdo who would enjoy listening to something like this though 😅

Here is an example for eth-nlped/stepverify

2 replies

·

posted an update 3 months ago

Post

2651

Hacked together a way to log trl GRPO training completions to a 🤗 dataset repo. This allows you to:

- Track rewards from multiple reward functions
- Treat the completion and rewards from training as a "proper" dataset and do EDA
- Share results for open science

The implementation is super hacky, but I'm curious if people would find this useful.

To push completions to the Hub, you just need two extra parameters:

log_completions=True
log_completions_hub_repo='your-username/repo-name'

Example dataset: davanstrien/test-logs
Colab: https://colab.research.google.com/drive/1wzBFPVthRYYTp-mEYlznLg_e_0Za1M3g

posted an update 4 months ago

Post

2282

Dataset descriptions for trending Hugging Face datasets? Powered by a Smol model davanstrien/Smol-Hub-tldr

posted an update 4 months ago

Post

1987

How do you make 1M+ Hugging Face models & datasets more discoverable?

davanstrien/Smol-Hub-tldr!

I fine-tuned HuggingFaceTB/SmolLM2-360M to generate one-line summaries from a model or dataset README.

Its own self-description?
"A model for generating concise summaries of model & dataset cards from the Hugging Face Hub"

The goal? Make it easier to find the right models and datasets for your specific needs. It's already powering a semantic search for datasets Space.

It's still a WIP but thanks to @loubnabnl , @anton-l , @eliebak et al, for cooking such a nice base model for fine-tuning small, efficient models for specific domains and tasks. 🙏

posted an update 4 months ago

Post

1408

Made some significant updates to my 🤗 semantic datasets search app. If you love falling into a wiki black hole, you might like this...

https://huggingface.co/spaces/librarian-bots/huggingface-datasets-semantic-search

reacted to Ihor's post with 🚀 4 months ago

Post

1692

🚀 Reproducing DeepSeek R1 for Text-to-Graph Extraction

I’ve been working on replicating DeepSeek R1, focusing on zero-shot text-to-graph extraction—a challenging task where LMs extract entities and relations from text based on predefined types.

🧠 Key Insight:
Language models struggle when constrained by entity/relation types. Supervised training alone isn’t enough, but reinforcement learning (RL), specifically Guided Reward Policy Optimization (GRPO), shows promise.

💡 Why GRPO?
It trains the model to generate structured graphs, optimizing multiple reward functions (format, JSON validity, and extraction accuracy).
It allows the model to learn from both positive and hard negative examples dynamically.
RL can be fine-tuned to emphasize relation extraction improvements.

📊 Early Results:
Even with limited training, F1 scores consistently improved, and we saw clear benefits from RL-based optimization. More training = better performance!

🔬 Next Steps:
We’re scaling up experiments with larger models and high-quality data. Stay tuned for updates! Meanwhile, check out one of our experimental models here:
Ihor/Text2Graph-R1-Qwen2.5-0.5b

📔 Learn more details from the blog post: https://medium.com/p/d8b648d9f419

Feel free to share your thoughts and ask questions!

2 replies

·

posted an update 4 months ago

Post

1868

Why choose between strong LLM reasoning and efficient models?

Use DeepSeek to generate high-quality training data, then distil that knowledge into ModernBERT answerdotai/ModernBERT-base for fast, efficient classification.

Blog post: https://danielvanstrien.xyz/posts/2025/deepseek/distil-deepseek-modernbert.html

posted an update 4 months ago

Post

1997

Updated the ColPali Query Generator Space davanstrien/ColPali-Query-Generator to use Qwen/Qwen2.5-VL-7B-Instruct.

Given an input image, it generates several queries along with explanations to justify them. This approach can generate synthetic data for fine-tuning ColPali models.

Daniel van Strien PRO

AI & ML interests

Recent Activity

Organizations

davanstrien's activity