Balancing Pipeline Parallelism with Vocabulary Parallelism
Paper
•
2411.05288
•
Published
•
18
Léo Hunout
hunoutl
AI & ML interests
AI Engineer working on Jean Zay supercomputer in France 🇫🇷
Organizations
hunoutl's activity
upvoted
14
papers
7 days ago
Face Anonymization Made Simple
Paper
•
2411.00762
•
Published
•
7
NeuZip: Memory-Efficient Training and Inference with Dynamic Compression of Neural Networks
Paper
•
2410.20650
•
Published
•
16
TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters
Paper
•
2410.23168
•
Published
•
22
Zipfian Whitening
Paper
•
2411.00680
•
Published
•
9
GPT or BERT: why not both?
Paper
•
2410.24159
•
Published
•
13
Sparsing Law: Towards Large Language Models with Greater Activation Sparsity
Paper
•
2411.02335
•
Published
•
11
Adaptive Caching for Faster Video Generation with Diffusion Transformers
Paper
•
2411.02397
•
Published
•
20
Self-Consistency Preference Optimization
Paper
•
2411.04109
•
Published
•
13
Polynomial Composition Activations: Unleashing the Dynamics of Large Language Models
Paper
•
2411.03884
•
Published
•
20
Animate-X: Universal Character Image Animation with Enhanced Motion Representation
Paper
•
2410.10306
•
Published
•
52
What Matters in Transformers? Not All Attention is Needed
Paper
•
2406.15786
•
Published
•
29
Neural Metamorphosis
Paper
•
2410.11878
•
Published
•
8
Movie Gen: A Cast of Media Foundation Models
Paper
•
2410.13720
•
Published
•
88
upvoted
3
papers
8 days ago
AutoTrain: No-code training for state-of-the-art models
Paper
•
2410.15735
•
Published
•
57
Mixture-of-Transformers: A Sparse and Scalable Architecture for Multi-Modal Foundation Models
Paper
•
2411.04996
•
Published
•
48
BitNet a4.8: 4-bit Activations for 1-bit LLMs
Paper
•
2411.04965
•
Published
•
63
upvoted
an
article
21 days ago
Article
Fine-tune Llama 3 with ORPO
By
•
•
227
upvoted
2
papers
about 2 months ago