FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling
Paper
•
2502.14856
•
Published
•
8
FR-Spec
Released ckpt for arxiv.org/abs/2502.14856
-
-
thunlp/LLaMA3-Instruct-8B-FR-Spec
Text Generation • Updated • 101 • 3 -
thunlp/Qwen2-7B-Instruct-FR-Spec
Text Generation • Updated • 12 • 1 -
thunlp/LLaMA3.2-Instruct-1B-FR-Spec
Text Generation • 0.3B • Updated • 29 • 2 -
openbmb/MiniCPM4-8B-Eagle-FRSpec
Text Generation • Updated • 92 • 8 -
openbmb/MiniCPM4-8B-Eagle-FRSpec-QAT-cpmcu
Text Generation • Updated • 80 • 10