πͺπΊβοΈ EU AI Act: Systemic Risks in the First CoP Draft Comments βοΈπͺπΊ
β’
11
Yacine Jernite
yjernite
AI & ML interests
Technical, community, and regulatory tools of AI governance @HuggingFace
Recent Activity
liked
a model
6 days ago
PatronusAI/glider
liked
a Space
6 days ago
PatronusAI/GLIDER
liked
a model
6 days ago
answerdotai/ModernBERT-large
Articles
Organizations
yjernite's activity
liked
a
model
6 days ago
liked
a
Space
6 days ago
liked
a
model
6 days ago
upvoted
a
collection
6 days ago
liked
a
Space
6 days ago
updated
a
dataset
7 days ago
liked
a
Space
7 days ago
reacted to
merve's
post with π
8 days ago
Post
2905
Apollo is a new family of open-source video language models by Meta, where 3B model outperforms most 7B models and 7B outperforms most 30B models π§Ά
β¨ the models come in 1.5B https://huggingface.co./Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co./Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co./Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
β¨ the authors also release a benchmark dataset https://huggingface.co./spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work β―οΈ
Try the demo for best setup here https://huggingface.co./spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled π scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful π₯
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models π₯
β¨ the models come in 1.5B https://huggingface.co./Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co./Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co./Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
β¨ the authors also release a benchmark dataset https://huggingface.co./spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work β―οΈ
Try the demo for best setup here https://huggingface.co./spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled π scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful π₯
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models π₯
- 3 replies
reacted to
fdaudens's
post with π
8 days ago
Post
1262
Did a fun experiment: What are the main themes emerging from the 100+ Nieman Journalism Lab predictions for 2025?
I used natural language processing to cluster and map them β really helps spot patterns that weren't obvious when reading predictions one by one. So what will shape journalism next year? A lot of AI and US politics (surprise!), but there's also this horizontal axis that spans from industry strategies to deep reflections on how to talk to the public.
Click any dot to explore the original prediction. What themes surprise/interest you the most?
π fdaudens/nieman_lab_2025_predictions_visualization
P.s.: I discovered that Nieman Lab's content is under Creative Commons license!
I used natural language processing to cluster and map them β really helps spot patterns that weren't obvious when reading predictions one by one. So what will shape journalism next year? A lot of AI and US politics (surprise!), but there's also this horizontal axis that spans from industry strategies to deep reflections on how to talk to the public.
Click any dot to explore the original prediction. What themes surprise/interest you the most?
π fdaudens/nieman_lab_2025_predictions_visualization
P.s.: I discovered that Nieman Lab's content is under Creative Commons license!
liked
a
Space
8 days ago
upvoted
an
article
8 days ago
Article
Finding Moroccan Arabic (Darija) in Fineweb 2
By
β’
β’
19
liked
a
Space
13 days ago
posted
an
update
13 days ago
Post
2035
πͺπΊ Policy Thoughts in the EU AI Act Implementation πͺπΊ
There is a lot to like in the first draft of the EU GPAI Code of Practice, especially as regards transparency requirements. The Systemic Risks part, on the other hand, is concerning for both smaller developers and for external stakeholders.
I wrote more on this topic ahead of the next draft. TLDR: more attention to immediate large-scale risks and to collaborative solutions supported by evidence can help everyone - as long as developers disclose sufficient information about their design choices and deployment contexts.
Full blog here, based on our submitted response with @frimelle and @brunatrevelin :
https://huggingface.co./blog/yjernite/eu-draft-cop-risks#on-the-proposed-taxonomy-of-systemic-risks
There is a lot to like in the first draft of the EU GPAI Code of Practice, especially as regards transparency requirements. The Systemic Risks part, on the other hand, is concerning for both smaller developers and for external stakeholders.
I wrote more on this topic ahead of the next draft. TLDR: more attention to immediate large-scale risks and to collaborative solutions supported by evidence can help everyone - as long as developers disclose sufficient information about their design choices and deployment contexts.
Full blog here, based on our submitted response with @frimelle and @brunatrevelin :
https://huggingface.co./blog/yjernite/eu-draft-cop-risks#on-the-proposed-taxonomy-of-systemic-risks
- 2 replies
upvoted
an
article
13 days ago
Article
πͺπΊβοΈ EU AI Act: Systemic Risks in the First CoP Draft Comments βοΈπͺπΊ
By
β’
β’
11
reacted to
dvilasuero's
post with β€οΈπ₯
19 days ago
Post
2259
π Announcing Global-MMLU: an improved MMLU Open dataset with evaluation coverage across 42 languages, built with Argilla and the Hugging Face community.
Global-MMLU is the result of months of work with the goal of advancing Multilingual LLM evaluation. It's been an amazing open science effort with collaborators from Cohere For AI, Mila - Quebec Artificial Intelligence Institute, EPFL, Massachusetts Institute of Technology, AI Singapore, National University of Singapore, KAIST, Instituto Superior TΓ©cnico, Carnegie Mellon University, CONICET, and University of Buenos Aires.
π·οΈ +200 contributors used Argilla MMLU questions where regional, dialect, or cultural knowledge was required to answer correctly. 85% of the questions required Western-centric knowledge!
Thanks to this annotation process, the open dataset contains two subsets:
1. π½ Culturally Agnostic: no specific regional, cultural knowledge is required.
2. βοΈ Culturally Sensitive: requires dialect, cultural knowledge or geographic knowledge to answer correctly.
Moreover, we provide high quality translations of 25 out of 42 languages, thanks again to the community and professional annotators leveraging Argilla on the Hub.
I hope this will ensure a better understanding of the limitations and challenges for making open AI useful for many languages.
Dataset: CohereForAI/Global-MMLU
Global-MMLU is the result of months of work with the goal of advancing Multilingual LLM evaluation. It's been an amazing open science effort with collaborators from Cohere For AI, Mila - Quebec Artificial Intelligence Institute, EPFL, Massachusetts Institute of Technology, AI Singapore, National University of Singapore, KAIST, Instituto Superior TΓ©cnico, Carnegie Mellon University, CONICET, and University of Buenos Aires.
π·οΈ +200 contributors used Argilla MMLU questions where regional, dialect, or cultural knowledge was required to answer correctly. 85% of the questions required Western-centric knowledge!
Thanks to this annotation process, the open dataset contains two subsets:
1. π½ Culturally Agnostic: no specific regional, cultural knowledge is required.
2. βοΈ Culturally Sensitive: requires dialect, cultural knowledge or geographic knowledge to answer correctly.
Moreover, we provide high quality translations of 25 out of 42 languages, thanks again to the community and professional annotators leveraging Argilla on the Hub.
I hope this will ensure a better understanding of the limitations and challenges for making open AI useful for many languages.
Dataset: CohereForAI/Global-MMLU
reacted to
fdaudens's
post with β€οΈ
21 days ago
Post
1053
ππ Just dropped: visualization mapping Hugging Face's most liked & downloaded models from 2022 to now. Small models are clearly on the rise - fascinating shift in both likes and download patterns.
Check it out: huggingface/open-source-ai-year-in-review-2024
Check it out: huggingface/open-source-ai-year-in-review-2024
reacted to
AdinaY's
post with β€οΈ
21 days ago
Post
1465
2023 & 2024 Top Downloaded (all time) Open Models on the hub are both from the Chinese community π
2023 π Bge base by BAAI
BAAI/bge-base-en-v1.5
2024 π Qwen 2.5 by Alibaba Qwen
Qwen/Qwen2.5-1.5B-Instruct
Canβt wait to see what incredible models the Chinese community will bring in 2025π
β¨ Follow https://huggingface.co./zh-ai-community to get the latest updates from the Chinese community
β¨ Explore the 2024 Year in Review huggingface/open-source-ai-year-in-review-2024
2023 π Bge base by BAAI
BAAI/bge-base-en-v1.5
2024 π Qwen 2.5 by Alibaba Qwen
Qwen/Qwen2.5-1.5B-Instruct
Canβt wait to see what incredible models the Chinese community will bring in 2025π
β¨ Follow https://huggingface.co./zh-ai-community to get the latest updates from the Chinese community
β¨ Explore the 2024 Year in Review huggingface/open-source-ai-year-in-review-2024
liked
a
Space
22 days ago
liked
a
Space
23 days ago