 Oliver Simons
|
021cc28bef
cuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs (#14741)
|
6 месяцев назад |
 Georgi Gerganov
|
d498af3d5a
graph : avoid huge warm-up graphs for MoE models (#14753)
|
6 месяцев назад |
|
Georgi Gerganov
|
eacdeb5bfc
model : fix build after merge conflict (#14754)
|
6 месяцев назад |
 lgai-exaone
|
e0cb5c5cb8
model : add EXAONE 4.0 support (#14630)
|
6 месяцев назад |
 Aman Gupta
|
f9a31eea06
CUDA: set_rows + cpy.cu refactor (#14712)
|
6 месяцев назад |
|
Georgi Gerganov
|
8f974bc1e9
graph : refactor context to not pass gf explicitly (#14629)
|
6 месяцев назад |
 Nexes the Elder
|
09651d09ff
graph : Pass the graph placeholder message in debug mode (#14748)
|
6 месяцев назад |
 Neo Zhang Jianyu
|
349ea79fce
use max work group size for device to replace the magic number (#14732)
|
6 месяцев назад |
 Piotr Wilkin (ilintar)
|
670e1360cd
convert : fix Ernie4.5 MoE without shared experts (#14746)
|
6 месяцев назад |
 Wroclaw
|
760b4484e3
nix : use optionalAttrs for env mkDerivation attrset argument (#14726)
|
6 месяцев назад |
|
Piotr Wilkin (ilintar)
|
cb887f1bc1
model: add Ernie 4.5 MoE support (#14658)
|
6 месяцев назад |
|
Georgi Gerganov
|
d6fb3f6b49
kv-cache : fix k-shift for multiple streams (#14742)
|
6 месяцев назад |
|
Georgi Gerganov
|
01612b7409
llama : reuse compute graphs (#14482)
|
6 месяцев назад |
 Tarek Dakhran
|
086cf81e88
llama : fix parallel processing for lfm2 (#14705)
|
6 месяцев назад |
|
Georgi Gerganov
|
d9b691081c
kv-cache : opt mask set input (#14600)
|
6 месяцев назад |
|
Georgi Gerganov
|
ad57d3edd2
batch : fix uninitialized has_cpl flag (#14733)
|
6 месяцев назад |
 Sigbjørn Skjæret
|
1ba45d4982
ci : disable failing vulkan crossbuilds (#14723)
|
6 месяцев назад |
|
Sigbjørn Skjæret
|
19e5943d9e
convert : make hf token optional (#14717)
|
6 месяцев назад |
 Diner Burger
|
496957e1cb
llama : fix parameter order for hybrid memory initialization (#14725)
|
6 месяцев назад |
 Reese Levine
|
21c021745d
ggml: Add initial WebGPU backend (#14521)
|
6 месяцев назад |
 tempstudio
|
b0f0ecc3dc
model : support output bias for qwen2 (#14711)
|
6 месяцев назад |
|
Georgi Gerganov
|
225e7a1438
llama : add high-throughput mode (#14363)
|
6 месяцев назад |
|
Aman Gupta
|
ab14019821
Support diffusion models: Add Dream 7B (#14644)
|
6 месяцев назад |
|
Georgi Gerganov
|
64978340b0
ggml : add asserts (#14720)
|
6 месяцев назад |
|
Georgi Gerganov
|
6ffd4e9c44
server : pre-calculate EOG logit biases (#14721)
|
6 месяцев назад |
 Shunta Saito
|
e4841d24d3
llama : fix parallel processing for plamo2 (#14716)
|
6 месяцев назад |
|
Georgi Gerganov
|
538cc77f7f
server : fix handling of the ignore_eos flag (#14710)
|
6 месяцев назад |
 Johannes Gäßler
|
5cae766541
scripts: synthetic prompt mode for server-bench.py (#14695)
|
6 месяцев назад |
|
Sigbjørn Skjæret
|
4b91d6f71f
convert : only check for tokenizer folder if we need it (#14704)
|
6 месяцев назад |
|
Sigbjørn Skjæret
|
cf91f217f1
convert : add pre-computed hashes first to prevent order mishaps (#14701)
|
6 месяцев назад |
