/ChunkLlama

[ICML'24] Data and code for our paper "Training-Free Long-Context Scaling of Large Language Models"

Primary LanguagePythonApache License 2.0Apache-2.0

Watchers