A few technical insights on the new Llama vision models we’re releasing today
RESEARCH
-
Data Analysis Essentials in Machine Learning Practice
By
–
new sticker dropped from @eugeneyan
's paper club session today QT with war stories of not looking at data -
Meta trains lightweight Llama models with net zero emissions
By
–
Combined, training these lightweight Llama models required 916K GPU hours on NVIDIA H100-80GB. As part of Meta’s commitment starting in 2020, we’ve maintained net zero greenhouse gas emissions in our global operations and matched 100% of the electricity use from the training
-
Llama 1B & 3B Models Released, Quantized Variants Coming Soon
By
–
It’s still early days for these models and we’re excited for the innovations ahead. The Llama 1B & 3B weights being released today are based on BFloat16 numerics. Our teams are actively exploring quantized variants that will run even faster and we hope to share more on that soon.
-
New Models Outperform Gemma 2 and Phi 3.5 at Smaller Sizes
By
–
We’re seeing exciting performance from these new models, with results that outperform Gemma 2 2.6B and Phi 3.5-mini models on a range of tasks even at smaller sizes!
-
Llama Lightweight Models: Pruning and Distillation Techniques
By
–
These lightweight Llama models were pretrained on up to 9 trillion tokens. One of the keys for Llama 1B & 3B however was using pruning & distillation to build smaller and more performant models informed by powerful teacher models. Pruning enabled us to reduce the size of extant
-
Technical insights on lightweight Llama 3.2 1B and 3B models
By
–
A few technical insights on our lightweight Llama 3.2 1B & 3B models.
-
Inside the Mathematics That Powers Artificial Intelligence
By
–
Inside the maths that drives AI
#AI #AIio #BigData #ML #NLU #Futureofwork @IIoT_World @ipfconline1 @jblefevre60 @JimMarous @insurtechtalk @KirkDBorne @kuriharan @LouisSerge -
Vector LoRA technique test and evaluation results
By
–
Oooh I do like this vector LoRA! (my usual test )
-
Code Generation Evaluation Scores and HumanEval Benchmarks
By
–
Exciting stuff! I don't see any evals on code gen. Do you have humaneval scores or anything like that available?