Using structured weight pruning and knowledge distillation, the @NVIDIAAI research team refined Llama 3.1 8B into a new Llama-3.1-Minitron 4B. They're releasing the new models on @huggingface and shared a deep dive on how they did it https://
go.fb.me/b2h2c8
@aiatmeta
-
NVIDIA Refines Llama 3.1 8B into Minitron 4B Model
By
–
-
European AI Startups: Last Call for 4-Month Accelerator Program
By
–
Last call — applications close this Friday https://
go.fb.me/cl9rjs This 4-month AI accelerator is designed for European startups developing advanced AI solutions with open-source foundation models. Our aim is to accelerate the development and market entry of these projects. -
Open Source Powers Cortex Analyst Preview Launch
By
–
This is the power of open source enabling incredible new possibilities, congratulations on the preview launch of Cortex Analyst!
-
Belebele Benchmark: Parallel Reading Comprehension in 122 Languages
By
–
The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants https://
go.fb.me/irdq2z -
LM Transparency Tool: Interactive Transformer Language Model Analysis
By
–
LM Transparency Tool: Interactive Tool for Analyzing Transformer Language Models https://
go.fb.me/5ngnnp -
MuTox: Universal Multilingual Audio-Based Toxicity Detector
By
–
MuTox: Universal MUltilingual Audio-based TOXicity Dataset and Zero-shot Detector
https://go.fb.me/y7vrjg -
Meta AI researchers present papers at ACL 2024 NLP conference
By
–
Following #ACL2024NLP from your feed? Here are papers to know about from AI researchers at Meta
-
Open approach to trust and safety with Llama Guard 3
By
–
New video on our open approach to trust & safety with Llama Guard 3, Prompt Guard & more. This extended discussion dives deep into system-level safety and how developers can build safer experiences from the start. Watch the full video https://
youtube.com/watch?v=IvjLXG
R7-vM
… -
Meta Announces LLM Evaluation Research Grant Program
By
–
LLM Evaluations are an important area of work — today we're announcing a new LLM Evaluation Research Grant to foster further innovation in this area. Recipients will get $200K in funding to support this work. We're accepting proposals until September 6 https://
go.fb.me/eym3xq -
SAM 2: Meta’s Real-Time Unified Model for Video Object Segmentation
By
–
SAM 2 from Meta FAIR is the first unified model for real-time, promptable object segmentation in images & videos. Using the model in our web-based demo you can segment, track and apply effects to objects in video in just a few clicks.
— AI at Meta (@AIatMeta) 7 août 2024
Try SAM 2 ➡️ https://t.co/dhf6reyoEu pic.twitter.com/QJoVRDJ1Z8SAM 2 from Meta FAIR is the first unified model for real-time, promptable object segmentation in images & videos. Using the model in our web-based demo you can segment, track and apply effects to objects in video in just a few clicks. Try SAM 2 https://
go.fb.me/7w0i4c