Tags

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

b1086

edd4c148 · llama : more tokenizer fixes (#2810) · Aug 27, 2023
b1085

1591e2e5 · ggml : detect SSSE3 (#2825) · Aug 27, 2023
b1083

c1ac54b7 · server : add `/detokenize` endpoint (#2802) · Aug 27, 2023
b1081

c7d92e6d · llama : use Unicode Escape Sequence to replace encoded characters (#2814) · Aug 26, 2023
b1079

741ca7dd · llama : move #includes out of _GNU_SOURCE conditional (#2817) · Aug 26, 2023
b1078

72f895c9 · main : fix bug (penalize_nl=false doesn't work) + suppress warning on mingw (#1528) · Aug 26, 2023
b1077

50526f37 · llama : use std::abs in llama_sample_tail_free (#2800) · Aug 26, 2023
b1076

04f4b1eb · k-quants : remove unnecessary tensor shape restrictions (#2811) · Aug 26, 2023
b1075

75923754 · Better perplexity for 2- and 3-bit quantization for LLaMA-v2-70B (#2807) · Aug 26, 2023
b1074

771551a7 · Fix HellaSwag (#2805) · Aug 26, 2023
b1071

2ba83c86 · Fix spm whitespaces (#2806) · Aug 26, 2023
ci_cublas_linux-b1071-5562e3e

5562e3e6 · temporarily disable broken 512 build · Aug 26, 2023
b1069

232caf3c · llama : fix struct decl (#2790) · Aug 25, 2023
b1068

d046dcee · Faster perplexity computation (#2786) · Aug 25, 2023
b1067

c82742ac · llama : add llama_beam_search() (#2267) · Aug 25, 2023
b1065

154725c5 · llama-bench : add model sizes (#2771) · Aug 25, 2023
b1063

29674ab4 · server : display token probabilities in the UI (#2489) · Aug 25, 2023
b1060

6bbc598a · ROCm Port (#1087) · Aug 25, 2023
b1059

3f460a2b · cuda : add RoPE kernel for mode == 2 (NeoX) (#2760) · Aug 25, 2023
b1057

b91ad7f4 · ggml-alloc : enlarge size of parse_seq (#2776) · Aug 25, 2023

Prev
1
…
61
62
63
64
65
66
67
68
69
…
99
Next