 DannyDaemonic
|
e8422de39e
@vxiiduu's fix for PrefetchVirtualMemory (#2930)
|
пре 2 година |
 Johannes Gäßler
|
8afe228000
CUDA: mul_mat_q=true llama_context_params default (#2912)
|
пре 2 година |
 Kawrakow
|
e37e69dcc3
10X faster BPE tokenizer (#2876)
|
пре 2 година |
 xaedes
|
44c117f41e
train : mem usage and other improvements (#2439)
|
пре 2 година |
|
Johannes Gäßler
|
6b73ef1201
YAML result logging + preset script (#2657)
|
пре 2 година |
 grahameth
|
be475f60af
llama.cpp : fix wrong vsnprintf call in MS compiler (#2856)
|
пре 2 година |
 Georgi Gerganov
|
c10704d01e
llama : fix MPI threads (close #2827)
|
пре 2 година |
|
Kawrakow
|
463173a6c0
llama : speedup tokenization (#2831)
|
пре 2 година |
|
Georgi Gerganov
|
eaa13a48ff
falcon : fix CUDA inference by making K and Q contiguous (#2830)
|
пре 2 година |
|
Kawrakow
|
a6d1189fdd
k_quants tuning for Falcon-7b (#2816)
|
пре 2 година |
|
Georgi Gerganov
|
d0cee0d36d
gguf : add 64-bit support (GGUF v2) (#2821)
|
пре 2 година |
|
Georgi Gerganov
|
edd4c14817
llama : more tokenizer fixes (#2810)
|
пре 2 година |
 Przemysław Pawełczyk
|
1591e2e590
ggml : detect SSSE3 (#2825)
|
пре 2 година |
 Tim Miller
|
c7d92e6dfe
llama : use Unicode Escape Sequence to replace encoded characters (#2814)
|
пре 2 година |
 Cebtenzzre
|
741ca7dd1c
llama : move #includes out of _GNU_SOURCE conditional (#2817)
|
пре 2 година |
|
Cebtenzzre
|
50526f37eb
llama : use std::abs in llama_sample_tail_free (#2800)
|
пре 2 година |
|
Georgi Gerganov
|
04f4b1eb10
k-quants : remove unnecessary tensor shape restrictions (#2811)
|
пре 2 година |
|
Kawrakow
|
7592375403
Better perplexity for 2- and 3-bit quantization for LLaMA-v2-70B (#2807)
|
пре 2 година |
 klosax
|
2ba83c8685
Fix spm whitespaces (#2806)
|
пре 2 година |
 Matt Pulver
|
c82742ac9c
llama : add llama_beam_search() (#2267)
|
пре 2 година |
 slaren
|
154725c543
llama-bench : add model sizes (#2771)
|
пре 2 година |
 Henri Vasserman
|
6bbc598a63
ROCm Port (#1087)
|
пре 2 година |
|
Georgi Gerganov
|
3f460a2b72
cuda : add RoPE kernel for mode == 2 (NeoX) (#2760)
|
пре 2 година |
|
slaren
|
0d3094f0c7
gguf : add rope_freq_base parameter for CodeLlama (#2769)
|
пре 2 година |
 Shouzheng Liu
|
38b16dfca6
metal : bug-fix when enable ggml-alloc (#2757)
|
пре 2 година |
|
slaren
|
fea95c682d
fix convert.py for codellama, add llama 34B to the list of recognized models (#2768)
|
пре 2 година |
|
Georgi Gerganov
|
c3e53b421a
llama : escape all U+2581 in a string (#2750)
|
пре 2 година |
 Evan Jones
|
6e91a1b070
llama : fix grammar sometimes generating null char (#2756)
|
пре 2 година |
|
Georgi Gerganov
|
cf658adc83
llm : add Falcon support (#2717)
|
пре 2 година |
 Kerfuffle
|
777f42ba18
Improve handling of special tokens in GGML to GGUF converter (#2725)
|
пре 2 година |
