exo
exo copied to clipboard
fix bfloat16 stuff
-
As numpy doesn't support bfloat16, updated code to check if it is bf16, and if yes, then convert to float32.
-
also renamed the output_data and first_layer variables to have a better understanding what contains what( like output_data_mx and output_data_np).
- As numpy doesn't support bfloat16, updated code to check if it is bf16, and if yes, then convert to float32.
- also renamed the output_data and first_layer variables to have a better understanding what contains what( like output_data_mx and output_data_np).
Related to #799 and #811