 Georgi Gerganov
|
ec450d3bbf
metal : opt-in compile flag for BF16 (#10218)
|
1 anno fa |
|
Georgi Gerganov
|
695ad752b2
metal : improve clarity (minor) (#10171)
|
1 anno fa |
|
Georgi Gerganov
|
841f27abdb
metal : optimize FA kernels (#10171)
|
1 anno fa |
 Jhen-Jie Hong
|
d05b3127bd
swift : exclude ggml-metal-embed.metal (#10211)
|
1 anno fa |
 Xuan Son Nguyen
|
76c6e7f105
server : minor UI fix (#10207)
|
1 anno fa |
|
Xuan Son Nguyen
|
a71d81cf8c
server : revamp chat UI with vuejs and daisyui (#10175)
|
1 anno fa |
|
Georgi Gerganov
|
eec4d71737
scripts : add amx to sync-ggml.sh [no ci]
|
1 anno fa |
|
Georgi Gerganov
|
3b08828674
sync : ggml
|
1 anno fa |
|
Georgi Gerganov
|
a2c6fd747c
scripts : sync update
|
1 anno fa |
 Diego Devesa
|
97404c4a03
ggml : add ggml-cpu.h to the public headers (#10204)
|
1 anno fa |
 Faisal Zaghloul
|
60e17ce23c
Remove identical wte/etw logic for jais (#10203)
|
1 anno fa |
 wwoodsTM
|
5107e8cea3
DRY: Fixes clone functionality (#10192)
|
1 anno fa |
 snadampal
|
2319126a70
fix q4_0_8_8 format for corrupted tokens issue (#10198)
|
1 anno fa |
 Zhiyuan Li
|
3bcd40b3c5
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (#10133)
|
1 anno fa |
|
Georgi Gerganov
|
5c333e0140
metal : add BF16 support (#8439)
|
1 anno fa |
|
Georgi Gerganov
|
b11f9ba9b8
server : remove hack for extra parallel slot (#10187)
|
1 anno fa |
|
Diego Devesa
|
94d8cb8be1
metal : fix from ptr buffer name (#10189)
|
1 anno fa |
|
Georgi Gerganov
|
1dc04b2dee
ggml : adjust is_first_call init value (#10193)
|
1 anno fa |
|
Georgi Gerganov
|
a1eaf6a960
metal : add quantized FA support (#10149)
|
1 anno fa |
 Gabe Goodhart
|
b8deef0ec0
llama : add <|tool_call|> formatting to Granite template (#10177)
|
1 anno fa |
|
Diego Devesa
|
a9e8a9a030
ggml : fix arch check in bf16_to_fp32 (#10164)
|
1 anno fa |
 Eve
|
3407364776
Q6_K AVX improvements (#10118)
|
1 anno fa |
|
Diego Devesa
|
d5a409e57f
ggml : fix gelu tables initialization (#10172)
|
1 anno fa |
|
Diego Devesa
|
401558b7ba
ggml : fix q4xx mat mul, increase ggml_aligned_malloc alignment (#10167)
|
1 anno fa |
|
Xuan Son Nguyen
|
9e0ecfb697
server : clarify /slots endpoint, add is_processing (#10162)
|
1 anno fa |
|
snadampal
|
6a066b9978
fix build break on arm64 linux (#10166)
|
1 anno fa |
|
Diego Devesa
|
ea02c753eb
cuda : clear error after changing peer access (#10153)
|
1 anno fa |
|
Georgi Gerganov
|
05697f670b
metal : simplify f16 and f32 dequant kernels (#0)
|
1 anno fa |
|
Georgi Gerganov
|
f8e58135cf
metal : move dequantize templates to beginning of MSL source (#0)
|
1 anno fa |
 leo-pony
|
329ed914c9
CANN: adjust backend registry refactor. (#10158)
|
1 anno fa |
