Tags

Tags give the ability to mark specific points in history as being important

b2845

b83cc3f5 · llama : add Jina Embeddings architecture (#6826) · May 11, 2024
b2844

9cb317f7 · ggml : full ALiBi support (#7192) · May 11, 2024
b2843

e8496488 · llama-bench : add pp+tg test type (#7199) · May 10, 2024
b2842

18e43766 · metal : fix flash attention kernel requirements (#7169) · May 10, 2024
b2840

25c6e82e · llama : use n_vocab to differentiate between mistral 7B and llama3 8B (#7200) · May 10, 2024
b2839

4e388097 · Fix memory bug in grammar parser (#7194) · May 10, 2024
b2838

f89fe273 · Main+: optionally allow special tokens from user in interactive mode (#7097) · May 10, 2024
b2837

d11afd66 · llava : fix moondream support (#7163) · May 10, 2024
b2836

8c570c94 · Minor arithmetic improvement to mmvq wrapper kernel (#7172) · May 10, 2024
b2835

eaf4bd8b · eval-callback : fix conversion to float (#7184) · May 10, 2024
b2834

befddd0f · Vulkan Bugfixes and Improvements (#7084) · May 09, 2024
b2831

43248e55 · llama3 custom regex split (#6965) · May 09, 2024
b2830

a743d76a · CUDA: generalize FP16 fattn vec kernel (#7061) · May 09, 2024
b2828

fd9f92b1 · llama : update llama_timings.n_p_eval setting (#7160) · May 09, 2024
b2826

47345248 · opencl : alignment size converted from bits to bytes (#7090) · May 09, 2024
b2824

4426e298 · cmake : fix typo (#7151) · May 08, 2024
b2822

bc4bba36 · Introduction of CUDA Graphs to LLama.cpp (#6766) · May 08, 2024
b2821

c12452c7 · JSON: [key] -> .at(key), assert() -> GGML_ASSERT (#7143) · May 08, 2024
b2820

9da243b3 · Revert "llava : add support for moondream vision language model (#6899)" · May 08, 2024
b2818

26458af1 · metal : use `vm_allocate` instead of `posix_memalign` on macOS (#7078) · May 08, 2024

Prev
1
…
5
6
7
8
9
10
11
12
13
…
99
Next