 Lee
|
a9559bf77b
ggml : workaround for missing _mm256_setr_m128i in GCC < 8 in k_quants.c (#2405)
|
%!s(int64=2) %!d(string=hai) anos |
 eric8607242
|
ee1b497c98
llama : support more diverse tokenizers? (#2420)
|
%!s(int64=2) %!d(string=hai) anos |
 Georgi Gerganov
|
d73b8d48b4
examples : fix whitespace
|
%!s(int64=2) %!d(string=hai) anos |
 nhamanasu
|
34ae1caf7f
examples : server chat mode with llama2 (#2400)
|
%!s(int64=2) %!d(string=hai) anos |
 Weird Constructor
|
d91f3f0c55
readme : fix the description of the Tail free sampling (TFS) method (#2431)
|
%!s(int64=2) %!d(string=hai) anos |
 Rand Xie
|
65cdf34bdc
llama : use n_embd_gqa instead of n_embd to handle llama-2 70B (#2433)
|
%!s(int64=2) %!d(string=hai) anos |
 niansa/tuxifan
|
edcc7ae7d2
Obtaining LLaMA 2 instructions (#2308)
|
%!s(int64=2) %!d(string=hai) anos |
 mj-shifu
|
7c529cede6
convert.py : Update to support 70B HF format model files (#2427)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
1a941869cb
metal : disable graph concurrency optimization due to bug (#2413)
|
%!s(int64=2) %!d(string=hai) anos |
 slaren
|
b5472ea0ad
ggml : fix assert in ggml_set_unary_op (#2410)
|
%!s(int64=2) %!d(string=hai) anos |
 Cebtenzzre
|
6df1f5940f
make : build with -Wmissing-prototypes (#2394)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
5488fb789e
ggml : allocate graphs in a context (#2392)
|
%!s(int64=2) %!d(string=hai) anos |
 Kawrakow
|
eb542d3932
Add LLAMA_DEFAULT_RMS_EPS so we can change the default (#2384)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
07aaa0f63f
ggml : fix ggml_flash_attn to use op_params (#2387)
|
%!s(int64=2) %!d(string=hai) anos |
 ldwang
|
fce48caf9a
convert.py : support bpe tokenizer (#2228)
|
%!s(int64=2) %!d(string=hai) anos |
 Jiahao Li
|
875086bdb9
ggml : relax contiguous constraints in activation function (#2371)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
da1889834a
ggml : improve graph build time via hash table lookup (#2329)
|
%!s(int64=2) %!d(string=hai) anos |
 Hesen Peng
|
82552b7f54
build : fix line breaking error in build-info.sh (#2349)
|
%!s(int64=2) %!d(string=hai) anos |
 Xiao-Yong Jin
|
0c06204fb3
main : add `--in-prefix-bos` to prefix BOS to user inputs; keep EOS (#2304)
|
%!s(int64=2) %!d(string=hai) anos |
 Eve
|
1fed755b1f
ci : add non-AVX scalar build/test (#2356)
|
%!s(int64=2) %!d(string=hai) anos |
 katsu560
|
be2301bcda
k_quants : add AVX support to dot functions with QK_K as 64 (#2339)
|
%!s(int64=2) %!d(string=hai) anos |
 Shouzheng Liu
|
1aa18ef994
metal : concurrently dispatch commands (#2358)
|
%!s(int64=2) %!d(string=hai) anos |
|
Kawrakow
|
9a08eaf3c4
Another speed gain for Q4_0 and Q4_1 on Metal (#2375)
|
%!s(int64=2) %!d(string=hai) anos |
|
Kawrakow
|
129d844c87
Fix Q4_K and Q5_K for QK_K = 64 on CUDA (#2359)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
d5512b782b
server: add rms_norm_eps parameter (#2380)
|
%!s(int64=2) %!d(string=hai) anos |
 Henri Vasserman
|
c798308e3a
[Server] Escape HTML in webchat (#2368)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
41c674161f
make rms_norm_eps a parameter (#2374)
|
%!s(int64=2) %!d(string=hai) anos |
 Aarni Koskela
|
b3f138d058
Chat UI extras (#2366)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
5b2b2dc6ae
ggml : sync (unary ops refactor, static-correctness) (#2370)
|
%!s(int64=2) %!d(string=hai) anos |
|
Kawrakow
|
42f70cb2f6
Fix scalar version of Q5_K when QK_K = 64 (#2362)
|
%!s(int64=2) %!d(string=hai) anos |
