 Georgi Gerganov
|
66841fdb0e
ggml : multi-thread mul and diag_mask ops (#1428)
|
2 лет назад |
 Johannes Gäßler
|
905d87b70a
ggml : GPU-accelerated token generation (#1412)
|
2 лет назад |
 xaedes
|
f954edda93
ggml : implement backward pass for llama + small training-llama-from-scratch example (#1360)
|
2 лет назад |
|
Georgi Gerganov
|
f048af0230
ggml : sync alibi fix from ggml repo
|
2 лет назад |
 3ooabkhxtn
|
ac0cd259d5
Adding SSE instructions to ggml_vec_dot_q4_0_q8_0 (#1413)
|
2 лет назад |
|
Georgi Gerganov
|
b9fd7eee57
ggml : remove bit shuffling (#1405)
|
2 лет назад |
 Sami Farin
|
9f8dbc4787
use pause asm insn in busyloop to run the CPU (13600K) 10 °C cooler (#1314)
|
2 лет назад |
 swittk
|
1b0fd45465
ggml : Allow usage of CLBlast alongside Accelerate.framework (#1336)
|
2 лет назад |
 Ron Jailall
|
20fbf2a2a0
ggml : change immintrin.h to intrin.h for compatibility (#1307)
|
2 лет назад |
|
Georgi Gerganov
|
799fdc1b5d
ggml : vectorize Q8_0 quantization
|
2 лет назад |
|
Georgi Gerganov
|
5d5817ca60
ggml : fix 32-bit ARM
|
2 лет назад |
 Marvin Gießing
|
cc0bb7235c
ggml : fix ppc64le build error and make cmake detect Power processors (#1284)
|
2 лет назад |
 slaren
|
2d099e5193
ggml: add names to tensors (#1268)
|
2 лет назад |
 slaren
|
58b367c2d7
cuBLAS: refactor and optimize f16 mat mul performance (#1259)
|
2 лет назад |
 Kerfuffle
|
2bdc09646d
ggml : fix ggml_used_mem() (#1264)
|
2 лет назад |
|
Georgi Gerganov
|
7ff0dcd320
ggml : fix UB (int << 31)
|
2 лет назад |
|
Georgi Gerganov
|
6bc4400e67
ggml : add Q5 WASM SIMD + GGML_FTYPE
|
2 лет назад |
|
Georgi Gerganov
|
3e5aa8a1c4
ggml : fix labels for GGML_OP_ALIBI
|
2 лет назад |
|
Georgi Gerganov
|
c3ca7a5f05
ggml : fix 32-bit ARM NEON
|
2 лет назад |
|
Georgi Gerganov
|
e8c051611a
ggml : use vzip instead of vuzp for consistency
|
2 лет назад |
|
Georgi Gerganov
|
0b5a935099
ggml : fix visibility and unused warnings
|
2 лет назад |
|
Georgi Gerganov
|
ec728e44d7
ggml : fix #if for f32_f32 mul_mat (CLBlast) (#1229)
|
2 лет назад |
|
Georgi Gerganov
|
214b6a3570
ggml : adjust mul_mat_f16 work memory (#1226)
|
2 лет назад |
|
slaren
|
7fc50c051a
cuBLAS: use host pinned memory and dequantize while copying (#1207)
|
2 лет назад |
 Henri Vasserman
|
b1ee8f59b4
cuBLAS: non-contiguous tensor support (#1215)
|
2 лет назад |
 Stephan Walter
|
36d19a603b
Remove Q4_3 which is no better than Q5 (#1218)
|
2 лет назад |
|
Georgi Gerganov
|
55390bcaf2
ggml : sync ggml (ggml_alibi)
|
2 лет назад |
|
Georgi Gerganov
|
11d902364b
ggml : add helper debug printf in soft_max
|
2 лет назад |
 0cc4m
|
7296c961d9
ggml : add CLBlast support (#1164)
|
2 лет назад |
 Yann Follet
|
04aaae1d79
add avx2 for dot_q8_0_q8_0, 2x faster than scalar (#1211)
|
2 лет назад |
