A Frame is Worth One Token: Efficient Generative World Modeling with Delta Tokens
Paper ā¢ 2604.04913 ā¢ Published ā¢ 6
AI & ML interests
This organization is maintained by the transformers team at Hugging Face and contains checkpoints of segmentation models such as SamHQ.
Recent Activity
View all activity
nielsrĀ
submitted a
paper to Daily Papers 3 days ago
nielsrĀ
submitted a
paper to Daily Papers 9 days ago
nielsrĀ
submitted a
paper to Daily Papers 20 days ago
nielsrĀ
submitted a
paper to Daily Papers 24 days ago
nielsrĀ
submitted a
paper to Daily Papers 25 days ago
nielsrĀ
authored a
paper about 1 month ago
nielsrĀ
submitted 2
papers to Daily Papers about 2 months ago
nielsrĀ
submitted a
paper to Daily Papers 2 months ago
nielsrĀ
submitted a
paper to Daily Papers 3 months ago
nielsrĀ
submitted a
paper to Daily Papers 4 months ago
Post
3474
š New blog: Maintain the unmaintainable ā 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 š¤ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
š Inside the post:
ā One Model, One File: readability first ā you can still open a modeling file and see the full logic, top to bottom.
ā Modular Transformers: visible inheritance that cuts maintenance cost by ~15Ć while keeping models readable.
ā 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 ā transformers-community/Transformers-tenets
How do you stop a million-line library built by thousands of contributors from collapsing under its own weight?
At š¤ Transformers, we do it with explicit software-engineering tenets, principles that make the codebase hackable at scale.
š Inside the post:
ā One Model, One File: readability first ā you can still open a modeling file and see the full logic, top to bottom.
ā Modular Transformers: visible inheritance that cuts maintenance cost by ~15Ć while keeping models readable.
ā 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 ā transformers-community/Transformers-tenets
nielsrĀ
updated 3
models 11 months ago
nielsrĀ
updated a
collection 11 months ago