 Pavol Rusnak
|
6f79699286
build: add armv{6,7,8} support to cmake (#1251)
|
2 vuotta sitten |
 Stephan Walter
|
f0d70f147d
Various fixes to mat_mul benchmark (#1253)
|
2 vuotta sitten |
 Georgi Gerganov
|
214b6a3570
ggml : adjust mul_mat_f16 work memory (#1226)
|
2 vuotta sitten |
|
Georgi Gerganov
|
305eb5afd5
build : fix reference to old llama_util.h
|
2 vuotta sitten |
 slaren
|
7fc50c051a
cuBLAS: use host pinned memory and dequantize while copying (#1207)
|
2 vuotta sitten |
 0cc4m
|
7296c961d9
ggml : add CLBlast support (#1164)
|
2 vuotta sitten |
 Johannes Gäßler
|
92a6e13a31
Add Manjaro CUDA include and lib dirs to Makefile (#1212)
|
2 vuotta sitten |
|
slaren
|
e4cf982e0d
Fix cuda compilation (#1128)
|
2 vuotta sitten |
|
Georgi Gerganov
|
e4422e299c
ggml : better PERF prints + support "LLAMA_PERF=1 make"
|
2 vuotta sitten |
|
Georgi Gerganov
|
872c365a91
ggml : fix AVX build + update to new Q8_0 format
|
2 vuotta sitten |
|
slaren
|
50cb666b8a
Improve cuBLAS performance by using a memory pool (#1094)
|
2 vuotta sitten |
|
slaren
|
2005469ea1
Add Q4_3 support to cuBLAS (#1086)
|
2 vuotta sitten |
 源文雨
|
5addcb120c
fix: LLAMA_CUBLAS=1 undefined reference 'shm_open' (#1080)
|
2 vuotta sitten |
|
slaren
|
02d6988121
Improve cuBLAS performance by dequantizing on the GPU (#1065)
|
2 vuotta sitten |
|
Stephan Walter
|
f3d4edf504
ggml : Q4 cleanup - remove 4-bit dot product code (#1061)
|
2 vuotta sitten |
|
slaren
|
8944a13296
Add NVIDIA cuBLAS support (#1044)
|
2 vuotta sitten |
 Kawrakow
|
5ecff35151
Adding a simple program to measure speed of dot products (#1041)
|
2 vuotta sitten |
|
Georgi Gerganov
|
e95b6554b4
ggml : add Q8_0 quantization for intermediate results (#951)
|
2 vuotta sitten |
|
Stephan Walter
|
93265e988a
make : fix dependencies, use auto variables (#983)
|
2 vuotta sitten |
|
Georgi Gerganov
|
9190e8eac8
llama : merge llama_internal.h into llama.h
|
2 vuotta sitten |
 CRD716
|
8cda5c981d
fix whitespace (#944)
|
2 vuotta sitten |
 SebastianApel
|
95ea26f6e9
benchmark : add tool for timing q4_0 matrix multiplication (#653)
|
2 vuotta sitten |
 comex
|
f963b63afa
Rewrite loading code to try to satisfy everyone:
|
2 vuotta sitten |
 unbounded
|
62cfc54f77
Add quantize-stats command for testing quantization (#728)
|
2 vuotta sitten |
 bhubbb
|
698f7b5d63
make : add libllama.so target for llama-cpp-python (#797)
|
2 vuotta sitten |
 Ivan Stepanov
|
0c44427df1
make : missing host optimizations in CXXFLAGS (#763)
|
2 vuotta sitten |
 Fabian
|
c4f89d8d73
make : use -march=native -mtune=native on x86 (#609)
|
2 vuotta sitten |
 david raistrick
|
1f0414feec
make : fix darwin f16c flags check (#615)
|
2 vuotta sitten |
|
Stephan Walter
|
436e561931
all : be more strict about converting float to double (#458)
|
2 vuotta sitten |
 RJ Adriaansen
|
4b8efff0e3
Add embedding example to Makefile (#540)
|
2 vuotta sitten |
