154e0d75fc
* Merging mainline - WIP * Merging mainline - WIP AVX2 and CUDA appear to work. CUDA performance seems slightly (~1-2%) lower as it is so often the case with llama.cpp/ggml after some "improvements" have been made. * Merging mainline - fix Metal * Remove check --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com> |
||
|---|---|---|
| .. | ||
| requirements-all.txt | ||
| requirements-compare-llama-bench.txt | ||
| requirements-convert_hf_to_gguf_update.txt | ||
| requirements-convert_hf_to_gguf.txt | ||
| requirements-convert_legacy_llama.txt | ||
| requirements-convert_llama_ggml_to_gguf.txt | ||
| requirements-convert_lora_to_gguf.txt | ||
| requirements-pydantic.txt | ||
| requirements-test-tokenizer-random.txt | ||