[ICLR 2025] SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration
Primary LanguagePythonApache License 2.0Apache-2.0
No issues in this repository yet.