Make more ML improvements

- Fix UX issues with llama.com
- Do housekeeping on libm code
- Add more vectorization to GGML
- Get GGJT quantizer programs working well
- Have the quantizer keep the output layer as f16c
- Prefetching improves performance 15% if you use fewer threads
This commit is contained in:
Justine Tunney 2023-05-16 08:07:23 -07:00
parent 80db9de173
commit e7eb0b3070
No known key found for this signature in database
GPG key ID: BE714B4575D6E328
46 changed files with 340 additions and 289 deletions

View file

@ -72,6 +72,7 @@
"__builtin_extract_return_addr"
"__builtin_isnan"
"__builtin_signbit"
"__builtin_signbitf"
"__builtin_signbitl"
"__builtin_ffs"
"__builtin_ffsl"