llama.cpp
1d7db017 - Materialize input_scales also for FP8 when converting

Commit

4 days ago

Materialize input_scales also for FP8 when converting

References

#27512 - Quant: OCP FP8 E4M3 support

#28392 - Add FP8 KV-cache support

#28413 - CUDA: Add OCP FP8 support

#28898 - ggml: support quant scales for fp8 and nvfp4

#29791 - vulkan: add fp8 and scaled matmul support

Author

ORippler

ORippler

Committer

ORippler

ORippler

Parents

Loading