 Georgi Gerganov
|
3600cc2886
llama : use n_swa + n_ubatch cells for SWA cache (#13833)
|
7 months ago |
 igardev
|
c7e0a2054b
webui : Replace alert and confirm with custom modals. (#13711)
|
7 months ago |
|
Georgi Gerganov
|
3f55f781f1
llama : auto-batch preparation (#13845)
|
7 months ago |
 Xuan-Son Nguyen
|
51fa76f172
mtmd : drop `_shared` from `libmtmd` name, merge helpers into libmtmd (⚠️ breaking change) (#13917)
|
7 months ago |
|
Georgi Gerganov
|
12d0188c0d
kv-cache : refactor + add llama_memory_state_i (#13746)
|
7 months ago |
 Shawn yang
|
eb3949938e
CUDA: add a prop in ggml_cuda_device_infor for distinguish iGPU or dGPU in cuda (#13856) (#13895)
|
7 months ago |
 Johannes Gäßler
|
e562eece7c
CUDA: fix typo in FlashAttention code (#13926)
|
7 months ago |
 Diego Devesa
|
b47ab7b8e9
sched : avoid changing cur_copy when a graph is already allocated (#13922)
|
7 months ago |
|
Georgi Gerganov
|
dd665cc9d4
parallel : increase the variability of the prompt lengths (#13927)
|
7 months ago |
|
Diego Devesa
|
df0c0c7d02
cuda : prevent using split buffers with 3d/4d matrices (#13919)
|
7 months ago |
 Akarshan Biswas
|
b49a8ff96b
SYCL: Add mrope kernel (#13755)
|
7 months ago |
|
Georgi Gerganov
|
53f925074d
sync : vendor (#13901)
|
7 months ago |
 Sigbjørn Skjæret
|
db38704f01
convert : fix rwkv bos/eos token (#13844)
|
7 months ago |
|
Xuan-Son Nguyen
|
07e4351ce6
convert : allow partial update to the chkhsh pre-tokenizer list (#13847)
|
7 months ago |
 Đinh Trọng Huy
|
291f2b6913
llama : add support for DistilBert (#13907)
|
7 months ago |
 zhangkaihuo
|
2c90da4c7e
llama : use llm_build_granite for minicpm (#13911)
|
7 months ago |
 Christian Kastner
|
ec9e0301fe
cmake: Guard GGML_CPU_ALL_VARIANTS by architecture (#13890)
|
7 months ago |
|
Sigbjørn Skjæret
|
e83ba3e460
llama : add support for jina-reranker-v2 (#13900)
|
7 months ago |
|
Sigbjørn Skjæret
|
2b131621e6
gguf-py : add support for sub_type (in arrays) in GGUFWriter add_key_value method (#13561)
|
7 months ago |
 Yibo Cai
|
54a2c7a8cd
arm64: optimize q4_k_q8_k kernel with i8mm (#13886)
|
7 months ago |
|
Christian Kastner
|
21fcc21ad5
cmake: Factor out CPU architecture detection (#13883)
|
7 months ago |
 Vineel Abhinav
|
dd8ba93416
ggml: aarch64: Implement SVE F32 kernels for Mamba Sequential Scan Algorithm (#13882)
|
7 months ago |
|
Georgi Gerganov
|
66c92061f5
tests : remove json.hpp from a test (#13880)
|
7 months ago |
|
Sigbjørn Skjæret
|
5ca82fc1d7
convert : workaround for AutoConfig dummy labels (#13881)
|
7 months ago |
|
Sigbjørn Skjæret
|
6385b843a8
llama : add RobertaForSequenceClassification reranker support (#13875)
|
7 months ago |
|
Vineel Abhinav
|
1b8fb8152d
ggml: aarch64: Implement SVE F32 kernels for vector functions (#13843)
|
7 months ago |
 Beinsezii
|
53ae30640e
gguf-py : fix SafetensorRemote return on undefined size (< 0) (#13841)
|
7 months ago |
|
Xuan-Son Nguyen
|
763d06edb7
llama : fix KV shift for qwen2vl (#13870)
|
7 months ago |
|
Xuan-Son Nguyen
|
10961339b2
mtmd : move helpers to dedicated library (⚠️ breaking change) (#13866)
|
7 months ago |
 bandoti
|
d98f2a35fc
ci: disable LLAMA_CURL for Linux cross-builds (#13871)
|
7 months ago |
