SafeAILab/EAGLE Issues
Qwen2 inference
Updated 2Reproduction Bias
Closed 4How to use multi-GPU when gen data?
Updated 2can eagle support
Closed 6Is this expected behavior?
Closed 2Will smaller models be supported?
Closed 1Question about EAGLE-2
Closed 5Why qx = 1.0 in Eagle2?
Closed 2EAGLE-2 is slower than EAGlE-1
Closed 8ge_data_all_llama2chat
ClosedQuality loss in greedy mode.
Closed 6How to handle embedding layernorm
Updated 9About reproducing speedup ratio
Closed 1Question about alpha
Closed 3Training config for llama3
Closed 2training sometimes crashes down
Updated 1number of draft token
Closed 2Llama 3 support?
Closed 3Question about sampling
Closed 6new model
Updated 6