[Misc] Refactor linear layer weight loading; introduce BasevLLMParameter
and weight_loader_v2
#17501
Job | Run time |
---|---|
1m 27s | |
1m 28s | |
1m 29s | |
57s | |
56s | |
6m 17s |
BasevLLMParameter
and weight_loader_v2
#17501
Job | Run time |
---|---|
1m 27s | |
1m 28s | |
1m 29s | |
57s | |
56s | |
6m 17s |