| .. |
|
__init__.py
|
f8dfac6372
chore: attention refactor and upstream sync apr01 (#365)
|
há 10 meses atrás |
|
cpu_executor.py
|
ef733aee43
implement ExecuteModelData to reduce executor complexity
|
há 6 meses atrás |
|
distributed_gpu_executor.py
|
de62ceb18c
refactor: eliminate parallel worker per-step task scheduling overhead
|
há 6 meses atrás |
|
executor_base.py
|
de62ceb18c
refactor: eliminate parallel worker per-step task scheduling overhead
|
há 6 meses atrás |
|
gpu_executor.py
|
236be273e5
feat: tensor parallel speculative decoding (#554)
|
há 6 meses atrás |
|
multiproc_gpu_executor.py
|
5b0c11d190
support pipeline parallel pynccl groups
|
há 6 meses atrás |
|
multiproc_worker_utils.py
|
eaa06fdd14
fix some f-strings
|
há 6 meses atrás |
|
neuron_executor.py
|
ef733aee43
implement ExecuteModelData to reduce executor complexity
|
há 6 meses atrás |
|
ray_gpu_executor.py
|
de62ceb18c
refactor: eliminate parallel worker per-step task scheduling overhead
|
há 6 meses atrás |
|
ray_utils.py
|
c6a501f682
add multiprocessing executor; make ray optional
|
há 6 meses atrás |
AlpinDale
5b0c11d190
support pipeline parallel pynccl groups