 Georgi Gerganov
|
2b3389677a
ggml : refactor rope norm/neox (#7634)
|
hai 1 ano |
|
Georgi Gerganov
|
d48c88cbd5
ggml : remove ggml_flash_attn and ggml_flash_ff (#7463)
|
hai 1 ano |
 liuwei-git
|
201cc11afa
llama : add phi3 128K model support (#7225)
|
hai 1 ano |
 Austin
|
8b1b1f4982
train : add general name (#6752)
|
hai 1 ano |
 Steve Grubb
|
6e0438da3c
gguf : fix resource leaks (#6061)
|
hai 1 ano |
|
Georgi Gerganov
|
ab336a9d5e
code : normalize enum names (#5697)
|
hai 1 ano |
 Herman Semenov
|
5d3de51f97
ggml, common, examples, tests : fixed type arguments in printf (#5528)
|
hai 1 ano |
 Daniel Bevenius
|
263978904c
finetune : rename feed-forward tensors (w1/w2/w3) (#4839)
|
hai 1 ano |
|
Georgi Gerganov
|
3b169441df
sync : ggml (#5452)
|
hai 1 ano |
 Uzo Nweke
|
381ee19572
finetune : fix ggml_allocr lifetimes (tmp workaround) (#5033)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
afefa319f1
ggml : change ggml_scale to take a float instead of tensor (#4573)
|
%!s(int64=2) %!d(string=hai) anos |
 Hongyu Ouyang
|
81bc9214a3
train : fix #4227 (double free in examples/train-text-from-scratch/train-text-from-scratch.cpp) (#4351)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
4760e7cc0b
sync : ggml (backend v2) (#3912)
|
%!s(int64=2) %!d(string=hai) anos |
 cebtenzzre
|
898aeca90a
llama : implement YaRN RoPE scaling (#2268)
|
%!s(int64=2) %!d(string=hai) anos |
 slaren
|
a5e8c1d8c7
train-text-from-scratch : fix assert failure in ggml-alloc (#3618)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
bc34dd4f5b
train : fix KQ_pos allocation (#3392)
|
%!s(int64=2) %!d(string=hai) anos |
|
Cebtenzzre
|
bc39553c90
build : enable more non-default compiler warnings (#3200)
|
%!s(int64=2) %!d(string=hai) anos |
|
slaren
|
16bc66d947
llama.cpp : split llama_context_params into model and context params (#3301)
|
%!s(int64=2) %!d(string=hai) anos |
 xaedes
|
0e76a8992c
train : finetune LORA (#2632)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
ec893798b7
llama : custom attention mask + parallel decoding + no context swaps (#3228)
|
%!s(int64=2) %!d(string=hai) anos |
 goerch
|
b08e75baea
Fixing the last deviations from sentencepiece indicated by test-tokenizer-1 (#3170)
|
%!s(int64=2) %!d(string=hai) anos |
|
Cebtenzzre
|
00d62adb79
fix some warnings from gcc and clang-tidy (#3038)
|
%!s(int64=2) %!d(string=hai) anos |
|
xaedes
|
44c117f41e
train : mem usage and other improvements (#2439)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
edd4c14817
llama : more tokenizer fixes (#2810)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
ef3f333d37
ggml : sync latest (SAM + SD operators, CUDA alibi) (#2709)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
6381d4e110
gguf : new file format with flexible meta data (beta) (#2398)
|
%!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
|
41c674161f
make rms_norm_eps a parameter (#2374)
|
%!s(int64=2) %!d(string=hai) anos |
|
Georgi Gerganov
|
513f861953
ggml : fix rope args order + assert (#2054)
|
%!s(int64=2) %!d(string=hai) anos |
 Spencer Sutton
|
5bf2a27718
ggml : remove src0 and src1 from ggml_tensor and rename opt to src (#2178)
|
%!s(int64=2) %!d(string=hai) anos |
