 Georgi Gerganov
|
1f63e75f3b
metal : use less stack memory in FA kernel (#14088)
|
hai 7 meses |
|
Georgi Gerganov
|
40cbf571c9
kv-cache : fix shift and defrag logic (#14081)
|
hai 7 meses |
 Diego Devesa
|
7f4fbe5183
llama : allow building all tests on windows when not using shared libs (#13980)
|
hai 7 meses |
 xctan
|
f470bc36be
ggml-cpu : split arch-specific implementations (#13892)
|
hai 7 meses |
|
Diego Devesa
|
8f47e25f56
cuda : fix device sync on buffer clear (#14033)
|
hai 7 meses |
|
Georgi Gerganov
|
201b31dc2e
graph : fix geglu (#14077)
|
hai 7 meses |
 Xinpeng Dou
|
e21d2d4ae2
CANN: Simplify the environment variable setting(#13104)
|
hai 7 meses |
 R0CKSTAR
|
dc0623fddb
webui: fix sidebar being covered by main content (#14082)
|
hai 7 meses |
|
Georgi Gerganov
|
87d34b381d
server : fix LRU check (#14079)
|
hai 7 meses |
 Nicolò Scipione
|
b460d16ae8
sycl: Add reorder to Q6_K mmvq implementation (#13885)
|
hai 7 meses |
 Đinh Trọng Huy
|
91a8ee6a6f
add geglu activation function (#14074)
|
hai 7 meses |
 Yuanhao Ji
|
056eb74534
CANN: Enable labeler for Ascend NPU (#13914)
|
hai 7 meses |
|
Diego Devesa
|
247e5c6e44
cuda : fix buffer type check with integrated GPUs (#14069)
|
hai 7 meses |
 吴小白
|
5787b5da57
ci: add LoongArch cross-compile build (#13944)
|
hai 8 meses |
 Akarshan Biswas
|
228f34c9ce
SYCL: Implement few same quantized type copy kernels (#13739)
|
hai 8 meses |
 Sigbjørn Skjæret
|
0974ad7a7c
llama : fix llama_model_chat_template with template name (LLM_KV with suffix) (#14050)
|
hai 8 meses |
|
Georgi Gerganov
|
745aa5319b
llama : deprecate llama_kv_self_ API (#14030)
|
hai 8 meses |
|
Georgi Gerganov
|
487a5e0401
context : fix SWA-related warning for multiple sequences (#14045)
|
hai 8 meses |
|
Sigbjørn Skjæret
|
d17a809ef0
llama : support multiple classifier outputs and labels (#13940)
|
hai 8 meses |
|
Sigbjørn Skjæret
|
1caae7fc6c
gguf-py : add add_classifier_output_labels method to writer (#14031)
|
hai 8 meses |
 Masato Nakasaka
|
669c13e0f6
vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (#14001)
|
hai 8 meses |
 pockers21
|
146b88e8b3
ci: fix CUDA build failure on autodl cloud machines (#14005)
|
hai 8 meses |
|
Georgi Gerganov
|
7f37b6cf1e
memory : migrate from llama_kv_cache to more generic llama_memory (#14006)
|
hai 8 meses |
|
Diego Devesa
|
3a077146a4
llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (#14013)
|
hai 8 meses |
 Olexandr88
|
d01d112abb
readme : add badge (#13938)
|
hai 8 meses |
|
Sigbjørn Skjæret
|
9f47fa5792
vocab : warn about missing mask token (#14022)
|
hai 8 meses |
|
Georgi Gerganov
|
9e31bec4fd
context : fix pos_min initialization upon error decode (#14008)
|
hai 8 meses |
 Jeff Bolz
|
5a8ae3053c
vulkan: automatically deduce size of push constants (#13936)
|
hai 8 meses |
 Ervin Áron Tasnádi
|
0d3984424f
ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (#13813)
|
hai 8 meses |
|
Georgi Gerganov
|
3e63a58ef7
kv-cache : refactor the update/defrag mechanism (#13988)
|
hai 8 meses |
