Tags

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

b1179

92177210 · speculative : add grammar support (#2991) · Sep 05, 2023
b1177

e36ecdcc · build : on Mac OS enable Metal by default (#2901) · Sep 04, 2023
b1176

bd33e5ab · ggml-opencl : store GPU buffer in ggml_tensor::extra (#2994) · Sep 04, 2023
b1175

31035681 · llama-bench : make cpp file non-executable (#2999) · Sep 04, 2023
b1174

5b8530d8 · make : add speculative example (#3003) · Sep 04, 2023
b1173

e4386f41 · server : add a subtle loading animation to the edit box (#2466) · Sep 04, 2023
b1172

35195689 · 2x faster (rms) norm cuda kernels (3.7% e2e improvement) (#2985) · Sep 04, 2023
b1171

cf9b0848 · ggml-alloc : use virtual memory for measurement (#2973) · Sep 03, 2023
b1170

47068e51 · speculative : PoC for speeding-up inference via speculative sampling (#2926) · Sep 03, 2023
b1169

8f429fa5 · perplexity : fix ETA by warming up the model with an empty run · Sep 03, 2023
b1165

37301347 · llama : fix bpe tokenize from byte (#2889) · Sep 03, 2023
b1163

afc43d5f · cov : add Code Coverage and codecov.io integration (#2928) · Sep 03, 2023
b1162

6460f758 · opencl : fix a bug in ggml_cl_pool_malloc() for ggml_cl_mul_mat_f32() (#2955) · Sep 03, 2023
b1157

c42f0ec6 · examples : fix gpt-neox (#2943) · Sep 03, 2023
b1155

bc054af9 · make : support overriding CFLAGS/CXXFLAGS/CPPFLAGS/LDFLAGS (#2886) · Sep 03, 2023
b1154

3358c381 · logging: Fix creating empty file even when disabled (#2966) · Sep 02, 2023
b1151

21f3d1be · k-quants : fix build on armv7 (android only) (#2920) · Sep 02, 2023
b1150

571083f5 · server : avoid aniprompt in probabilities of final response (#2849) · Sep 02, 2023
b1149

f04d0028 · cuda : vsubss4 for older versions of ROCm/clang (#2942) · Sep 01, 2023
b1147

5d6f19f1 · Allow quantize to only copy tensors, some other improvements (#2931) · Sep 01, 2023

Prev
1
…
58
59
60
61
62
63
64
65
66
…
99
Next