Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking
Paper
β’ 2602.21196 β’ Published
β’ 3
AI & ML interests
None defined yet.
Recent Activity
View all activity
mryabΒ
submitted a paper to Daily Papers 5 days ago
ariG23498Β
authored a paper 4 months ago
ericmsmithΒ
authored 14 papers 5 months ago
Multiple-Attribute Text Style Transfer
Paper
β’ 1811.00552 β’ Published
Towards Empathetic Open-domain Conversation Models: a New Benchmark and Dataset
Paper
β’ 1811.00207 β’ Published
β’ 1
Recipes for building an open-domain chatbot
Paper
β’ 2004.13637 β’ Published
β’ 1
Can You Put it All Together: Evaluating Conversational Agents' Ability to Blend Skills
Paper
β’ 2004.08449 β’ Published
β’ 1
Multi-Modal Open-Domain Dialogue
Paper
β’ 2010.01082 β’ Published
ROBBIE: Robust Bias Evaluation of Large Generative Language Models
Paper
β’ 2311.18140 β’ Published
β’ 1
Improving Open Language Models by Learning from Organic Interactions
Paper
β’ 2306.04707 β’ Published
β’ 3
"I'm sorry to hear that": Finding New Biases in Language Models with a Holistic Descriptor Dataset
Paper
β’ 2205.09209 β’ Published
Perturbation Augmentation for Fairer NLP
Paper
β’ 2205.12586 β’ Published
BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage
Paper
β’ 2208.03188 β’ Published
β’ 1
The Llama 3 Herd of Models
Paper
β’ 2407.21783 β’ Published
β’ 117
Persistent Pre-Training Poisoning of LLMs
Paper
β’ 2410.13722 β’ Published
Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations
Paper
β’ 2411.10414 β’ Published
β’ 1
Large Reasoning Models Learn Better Alignment from Flawed Thinking
Paper
β’ 2510.00938 β’ Published
β’ 59
Post
3420
π New blog: Maintain the unmaintainable β 1M+ Python LOC, 400+ models
How do you stop a million-line library built by thousands of contributors from collapsing under its own weight?
At π€ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
π Inside the post:
β One Model, One File: readability first β you can still open a modeling file and see the full logic, top to bottom.
β Modular Transformers: visible inheritance that cuts maintenance cost by ~15Γ while keeping models readable.
β Config-Driven Performance: FlashAttention, tensor parallelism, and attention scheduling are config-level features, not rewrites.
Written with @lysandre ,@pcuenq and @yonigozlan , this is a deep dive into how Transformers stays fast, open, and maintainable.
Read it here β transformers-community/Transformers-tenets
How do you stop a million-line library built by thousands of contributors from collapsing under its own weight?
At π€ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
π Inside the post:
β One Model, One File: readability first β you can still open a modeling file and see the full logic, top to bottom.
β Modular Transformers: visible inheritance that cuts maintenance cost by ~15Γ while keeping models readable.
β Config-Driven Performance: FlashAttention, tensor parallelism, and attention scheduling are config-level features, not rewrites.
Written with @lysandre ,@pcuenq and @yonigozlan , this is a deep dive into how Transformers stays fast, open, and maintainable.
Read it here β transformers-community/Transformers-tenets
burtenshawΒ
authored a paper 5 months ago
Post
8032
We're kick-starting the process of Transformers v5, with @ArthurZ and @cyrilvallez !
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.
Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!
- 6 replies
Post
1751
New post is live!
This time we cover some major updates to
π€
This time we cover some major updates to
transformers.π€
-
- 2 replies