benchmark
Fix bf16 inputs in six models and cm3leon_generate with CUDA graphs
#2717
Open
Open
Fix bf16 inputs in six models and cm3leon_generate with CUDA graphs #2717
ConnorBaker
wants to merge
8 commits into
pytorch:main
from
ConnorBaker:fix-bf16-inputs-and-cm3leon-cudagraphs
Fix dcgan's misspelled example_inputs so precision casts reach them
eccfeb3b
Run phlippe_resnet on example_inputs so precision casts reach them
a62b6088
Keep lennard_jones's inputs in example_inputs so precision casts reac…
744cd13a
Expose nvidia_deeprecommender's inputs as example_inputs
5694be91
Expose tts_angular's inputs as example_inputs
f515e630
Implement set_module() for maml
b97a0b0b
Don't cache cm3leon_generate's causal mask on the module
b29324b0
Initialize cm3leon_generate's Linear and attention weights
0b8ae56d
Reviewers
No reviews
Assignees
No one assigned
Labels
cla signed
Milestone
No milestone
Login to write a write a comment.
Login via GitHub