BetaRare
tomaarsen/attention_sinks
Extend existing LLMs way beyond the original training length with constant memory usage, without retraining
- Owner
- tomaarsen (personal account)
- Stars
- Stars: 734
- Forks
- Forks: 43
- Language
- Python
- License
- Apache-2.0
- Created
- 2023
- Topics
- llmllmsnlppythontransformers
- Tier
- Beta (700 to 1,999 ★) What this tier means
More repos by tomaarsen
All 3 repos by tomaarsen- tomaarsen/SpanMarkerNERStars: 475Jupyter NotebookAlpha
- tomaarsen/TwitchMarkovChainStars: 139PythonPrototype
Nearby in Python
All Python repos- tracefinity/tracefinityStars: 734PythonBeta
- yzhao062/agent-styleStars: 734PythonBeta
- ARahim3/mlx-dsparkStars: 734PythonBeta
- itdxer/neupyStars: 734PythonBeta
- czheo/syntax_sugar_pythonStars: 734PythonBeta
- Cadene/vqa.pytorchStars: 734PythonBeta