Tags

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

b1782

72d8407b · llama.swiftui : use llama.cpp as SPM package (#4804) · Jan 07, 2024
b1781

d117d4dc · llama : print tensor meta for debugging · Jan 07, 2024
b1779

63ee677e · ggml : use __builtin_amdgcn_sudot4 in __dp4a for gfx11 (#4787) · Jan 07, 2024
b1778

67984921 · server : fix n_predict check (#4798) · Jan 07, 2024
b1777

c75ca5d9 · llama.swiftui : use correct pointer for llama_token_eos (#4797) · Jan 06, 2024
b1775

eec22a1c · cmake : check for openblas64 (#4134) · Jan 05, 2024
b1773

91d38876 · metal : switch back to default.metallib (ggml/681) · Jan 05, 2024
b1770

c1d7cb28 · ggml : do not sched_yield when calling BLAS (#4761) · Jan 05, 2024
b1768

b3a7c20b · finetune : remove unused includes (#4756) · Jan 04, 2024
b1767

012cf349 · server : send token probs for "stream == false" (#4714) · Jan 04, 2024
b1766

a9192801 · Print backend name on test-backend-ops failure (#4751) · Jan 04, 2024
b1765

3c0b5855 · llama.swiftui : support loading custom model from file picker (#4767) · Jan 04, 2024
b1763

dc891b7f · ggml : include stdlib.h before intrin.h (#4736) · Jan 04, 2024
b1761

cb1e2818 · train : fix typo in overlapping-samples help msg (#4758) · Jan 03, 2024
b1760

ece9a45e · swift : update Package.swift to use ggml as dependency (#4691) · Jan 03, 2024
b1759

7bed7eba · cuda : simplify expression · Jan 03, 2024
b1752

f3f62f0d · metal : optimize ggml_mul_mat_id (faster Mixtral PP) (#4725) · Jan 02, 2024
b1751

0ef3ca2a · server : add token counts to html footer (#4738) · Jan 02, 2024
b1750

540938f8 · llama : llama_model_desc print number of experts · Jan 02, 2024
b1749

0040d42e · llama : replace all API facing `int`'s with `int32_t` (#4577) · Jan 02, 2024

Prev
1
…
36
37
38
39
40
41
42
43
44
…
99
Next