llama : require first token to be BOS (ggml-org#1303)

* llama : require first token to be BOS

* scripts : add ppl-run-all.sh

* perplexity : add BOS for each chunk

* readme : update perplexity values after BOS fix

* perplexity : add clarifying comments

llama : fix hparams shadow (ggml-org#1367)

fixes ggml-org#1363

Documented CUDA reproducibility, added warning (ggml-org#1346)

CI: add Windows CLBlast and OpenBLAS builds (ggml-org#1277)

* Add OpenCL and CLBlast support

* Add OpenBLAS support

* Remove testing from matrix

* change build name to 'clblast'

ggml : Allow usage of CLBlast alongside Accelerate.framework (ggml-or…

…g#1336)

Minor edit in ggml.c which originally would prevent OpenCL from loading completely if GGML_USE_ACCELERATE was defined.
Minor speedup in prompt eval time.

Remove default arguments from sampling functions (ggml-org#1343)

ci : add cublas to windows release (ggml-org#1271)

makefile: automatic Arch Linux detection (ggml-org#1332)

This commit is a port of a detection method used in koboldcpp's Makefile in order to automatically set the -lcblas option on Arch Linux

Fix for OpenCL / clbast builds on macOS. (ggml-org#1329)

fix ggml-org#1224 reverse prompt and multi line (ggml-org#1297)

* fix reverse prompt and multi line

* Code Formatting

Co-authored-by: Georgi Gerganov <[email protected]>

---------

Co-authored-by: Georgi Gerganov <[email protected]>