common: fix warning (ggml-org#8036)

* common: fix warning

* Update common/common.cpp

Co-authored-by: slaren <[email protected]>

---------

Co-authored-by: slaren <[email protected]>

[SYCL] Fix windows build and inference (ggml-org#8003)

* add sycl preset

* fix debug link error. fix windows crash

* update README

CUDA: stream-k decomposition for MMQ (ggml-org#8018)

* CUDA: stream-k decomposition for MMQ

* fix undefined memory reads for small matrices

metal : fix `ggml_metal_supports_op` for BF16 (ggml-org#8021)

Currently the Metal backend does not support BF16. `ggml_metal_supports_op` was returning true in these cases, leading to a crash with models converted with `--leave-output-tensor`. This commit checks if the first few sources types are BF16 and returns false if that's the case.

server : fix smart slot selection (ggml-org#8020)

ggml : synchronize threads using barriers (ggml-org#7993)

codecov : remove (ggml-org#8004)

[SYCL] refactor (ggml-org#6408)

* seperate lower precision GEMM from the main files

* fix workgroup size hardcode

tokenizer : BPE fixes (ggml-org#7530)

* Random test: add_bos_token, add_eos_token
* Random test: add BPE models for testing
* Custom regex split fails with codepoint 0
* Fix falcon punctuation regex
* Refactor llm_tokenizer_bpe: move code to constructor
* Move 'add_special_bos/eos' logic to llm_tokenizer_bpe
* Move tokenizer flags to vocab structure.
* Default values for special_add_bos/eos
* Build vocab.special_tokens_cache using vocab token types
* Generalize 'jina-v2' per token attributes
* Fix unicode whitespaces (deepseek-coder, deepseek-llm)
* Skip missing byte tokens (falcon)
* Better unicode data generation
* Replace char32_t with uint32_t

Only use FIM middle token if it exists (ggml-org#7648)

* Only use FIM middle if it exists

* Only use FIM middle if it exists