[Misc] Refactor linear layer weight loading; introduce BasevLLMParameter and weight_loader_v2 (#5874)

This commit is contained in:
Dipika Sikka
2024-08-07 12:17:58 -04:00
committed by GitHub
parent 639159b2a6
commit 0f7052bc7e
11 changed files with 653 additions and 201 deletions

View File

@@ -1,7 +1,11 @@
from vllm.model_executor.parameter import (BasevLLMParameter,
PackedvLLMParameter)
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.model_executor.utils import set_random_seed
__all__ = [
"SamplingMetadata",
"set_random_seed",
"BasevLLMParameter",
"PackedvLLMParameter",
]