llama.cpp
2075a66a - metal : fix `ggml_metal_supports_op` for BF16 (#8021)

Commit
1 year ago
metal : fix `ggml_metal_supports_op` for BF16 (#8021) Currently the Metal backend does not support BF16. `ggml_metal_supports_op` was returning true in these cases, leading to a crash with models converted with `--leave-output-tensor`. This commit checks if the first few sources types are BF16 and returns false if that's the case.
Author
Parents
Loading