Lorenzo Liusso

autocast bfloat16 to float32 for llama on clang

3

#6909 and #6905 The predefined function fix_bf16 was not working correctly. I was able to load the weights and run examples/llama3.py.

clang backend fails using version 18.1.3

5

using CLANG backend Model already downloaded using LLaMA-7B model 0%| | 0/292 [00:00