* llama : add thread safety test * llamafile : remove global state * llama : better LLAMA_SPLIT_MODE_NONE logic when main_gpu < 0 GPU devices are not used --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> |
||
|---|---|---|
| .. | ||
| ISSUE_TEMPLATE | ||
| actions | ||
| workflows | ||
| labeler.yml | ||
| pull_request_template.md | ||