scripts : update sync
62eeaafc
metal : fix minor string leaks (ggml/1004)
a53ac6f9
cmake : make it possible linking ggml as external lib (ggml/1003)
787b66fc
musa: workaround for Guilty Lockup in cleaning src0 (llama/10042)
85c678cf
ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the…
5d22f5b5
llama : refactor model loader with backend registry (llama/10026)
a0ea7d47
ggml : add Q4_0_8_8 RISC-V GEMV and GEMM kernels (llama/10029)
39fdd62f
ggml : fix memory leaks when loading invalid gguf files (llama/10094)
4cbca543
kompute: add backend registry / device interfaces (llama/10045)
d378f19e
kompute: add mul_mat_q4_k shader (llama/10097)
18122841
ggml : check tensor name lengths in gguf files (llama/10100)
1c837521
llama : fix buffer checks for mamba and rwk (llama/10111)
6b7f6bec
build: fix build error in Windows env with OneAPI setup (llama/10107)
72cbb25c
ggml : remove ggml_scratch (llama/10121)
6352fcd6
vulkan : improve ggml_vk_create_buffer error handling (llama/9898)
c28c6e85
llama : use smart pointers for ggml resources (llama/10117)
c7c5a952
llama : add simple-chat example (llama/10124)
749d287c
metal : minor fixup in FA kernel (llama/10143)
384ee00d
ggml : move CPU backend to a separate file (llama/10144)
63f72869
CANN: adjust backend registry refactor. (llama/10158)
fa240b25
metal : move dequantize templates to beginning of MSL source (llama/0)
e75a4534
metal : simplify f16 and f32 dequant kernels (llama/0)
e72fc8ab
cuda : clear error after changing peer access (llama/10153)
801fdc22
fix build break on arm64 linux (llama/10166)
03b75f41
ggml : fix q4xx mat mul, increase ggml_aligned_malloc alignment (llam…
c7655fe2
ggml : fix gelu tables initialization (llama/10172)
45ecfd9a
Q6_K AVX improvements (llama/10118)
26e5a758
ggml : fix arch check in bf16_to_fp32 (llama/10164)
7580d7ea
metal : add quantized FA support (llama/10149)
8aaa1e95
ggml : adjust is_first_call init value (llama/10193)
354191fb
metal : fix from ptr buffer name (llama/10189)
406c648a
metal : add BF16 support (llama/8439)
42cbecec
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acc…
b8fbee45
fix q4_0_8_8 format for corrupted tokens issue (llama/10198)
3cae70b9
ggml : add ggml-cpu.h to the public headers (llama/10204)
44c0abf3
metal : optimize FA kernels (llama/10171)
d6fb23bb
metal : improve clarity (minor) (llama/10171)
2b11c93f
metal : opt-in compile flag for BF16 (llama/10218)
6998ecfe
ggml : optimize llamafile cpu matrix multiplication for ppc64le (llam…
f48abc2b
ggml: fix zero division in ‘dne’ calculation in CUDA COUNT_EQUAL oper…
d54b0d2b
metal : hide debug messages from normal log
065fc31a
metal : fix F32 accumulation in FA vec kernel (llama/10232)
be6999ed
metal : fix build and some more comments (llama/10229)
e47d0eb0
metal : reorder write loop in mul mat kernel + style (llama/10231)
8536022f
vulkan: Fix newly added tests for permuted mul_mat and 1D im2col (lla…
db5507a3
metal : more precise Q*K in FA vec kernel (llama/10247)
c4c4d888
vulkan: Throttle the number of shader compiles during the build step.…
b606ad2f
vulkan: Optimize contiguous copies (llama/10254)
c60ae68c
sycl : Fixes to broken builds and test-backend-ops (llama/10257)
3b637d42
sync : ggml
3c337b22
whisper : fix build (#0)
3df5e16d
talk-llama : sync llama.cpp
d93631c4
build : fixes
19927ad8
whisper : include ggml-cpu.h (#0)
463849af
cmake : fix ppc64 check (#0)
f94863ee
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub