Ali El Filali's picture

Ali El Filali

alielfilali01

AI & ML interests

AI Psychometrician ? | NLP (mainly for Arabic) | Other interests include Reinforcement Learning and Cognitive sciences among others

Recent Activity

updated a dataset 1 day ago
inceptionai/requests-dataset
upvoted a collection 2 days ago
Deepseek Papers
upvoted a paper 2 days ago
DeepSeek-V3 Technical Report
View all activity

Articles

Organizations

Gradio-Themes-Party's profile picture Arabic Machine Learning 's profile picture BigLAM: BigScience Libraries, Archives and Museums's profile picture Stable Diffusion Dreambooth Concepts Library's profile picture Blog-explorers's profile picture ASAS AI's profile picture Nt3awnou's profile picture Qwen's profile picture Mixed Arabic Datasets's profile picture ZeroGPU Explorers's profile picture 2A2I Legacy Models & Datasets's profile picture AtlasIA's profile picture 2A2I's profile picture Open Arabic LLM Leaderboard's profile picture MLX Community's profile picture Social Post Explorers's profile picture C4AI Community's profile picture Dev Mode Explorers's profile picture Chinese LLMs on Hugging Face's profile picture ThinkAI's profile picture KABOUR's profile picture Hugging Face Discord Community's profile picture llmc's profile picture Arabic Translation Prompt Engineering's profile picture Inception's profile picture Dataset Tools's profile picture ml-fw-prerelease's profile picture Data Is Better Together Contributor's profile picture Donut Earthers 🍩's profile picture QudraTech's profile picture

Posts 28

view post
Post
1542
~75% on the challenging GPQA with only 40M parameters 🔥🥳

GREAT ACHIEVEMENT ! Or is it ?

This new Work, "Data Laundering: Artificially Boosting Benchmark Results through Knowledge Distillation", take out the mystery about many models i personally suspected their results. Speacially on leaderboards other than the english one, Like the Open Arabic LLM Leaderbaord OALL/Open-Arabic-LLM-Leaderboard.

The authors of this work, first started by training a model on the GPQA data, which, unsurprisingly, led to the model achieving 100% performance.

Afterward, they trained what they referred to as a 'legitimate' model on legitimate data (MedMCQA). However, they introduced a distillation loss from the earlier, 'cheated' model.

What they discovered was fascinating: the knowledge of GPQA leaked through this distillation loss, even though the legitimate model was never explicitly trained on GPQA during this stage.

This raises important questions about the careful use of distillation in model training, especially when the training data is opaque. As they demonstrated, it’s apparently possible to (intentionally or unintentionally) leak test data through this method.

Find out more: Data Laundering: Artificially Boosting Benchmark Results through Knowledge Distillation (2412.15255)
view post
Post
3353
Unpopular opinion: Open Source takes courage to do !

Not everyone is brave enough to release what they have done (the way they've done it) to the wild to be judged !
It really requires a high level of "knowing wth are you doing" ! It's kind of a super power !

Cheers to the heroes here who see this!