 Neo Zhang Jianyu
|
95bc82fbc0
[SYCL] add missed dll file in package (#9577)
|
hai 1 ano |
 R0CKSTAR
|
7691654c68
mtgpu: enable VMM (#9597)
|
hai 1 ano |
 Xuan Son Nguyen
|
ea9c32be71
ci : fix docker build number and tag name (#9638)
|
hai 1 ano |
 Charles Xu
|
1e43630218
ggml : remove assert for AArch64 GEMV and GEMM Q4 kernels (#9217)
|
hai 1 ano |
|
Xuan Son Nguyen
|
afbbfaa537
server : add more env vars, improve gen-docs (#9635)
|
hai 1 ano |
 Gabe Goodhart
|
3d6bf6919f
llama : add IBM Granite MoE architecture (#9438)
|
hai 1 ano |
 Dou Xinpeng
|
904837e0cb
cann: fix crash when llama-bench is running on multiple cann devices (#9627)
|
hai 1 ano |
 Eric Zhang
|
70392f1f81
ggml : add AVX512DQ requirement for AVX512 builds (#9622)
|
hai 1 ano |
 Georgi Gerganov
|
bb5f819975
sync : ggml
|
hai 1 ano |
|
Georgi Gerganov
|
c038931615
examples : adapt to ggml.h changes (ggml/0)
|
hai 1 ano |
|
Georgi Gerganov
|
31ac5834fe
llama : keep track of all EOG tokens in the vocab (#9609)
|
hai 1 ano |
|
Georgi Gerganov
|
cea1486ecf
log : add CONT level for continuing previous log entry (#9610)
|
hai 1 ano |
 StrangeBytesDev
|
0aa15011e3
server : add newline after chat example (#9616)
|
hai 1 ano |
|
Georgi Gerganov
|
b0f27361f3
sampling : avoid expensive softmax during greedy sampling (#9605)
|
hai 1 ano |
 Max Krasnyansky
|
c087b6f11d
threads: fix msvc build without openmp (#9615)
|
hai 1 ano |
 Ivan
|
116efee0ee
cuda: add q8_0->f32 cpy operation (#9571)
|
hai 1 ano |
|
Xuan Son Nguyen
|
0b3bf966f4
server : add --no-context-shift option (#9607)
|
hai 1 ano |
|
Max Krasnyansky
|
f0c7b5edf8
threads: improve ggml_barrier scaling with large number of threads (#9598)
|
hai 1 ano |
 Riceball LEE
|
1d48e98e4f
readme : add programmable prompt engine language CLI (#9599)
|
hai 1 ano |
|
Georgi Gerganov
|
f3979df762
flake.lock: Update (#9586)
|
hai 1 ano |
 Srihari-mcw
|
1e7b9299c6
ggml : AVX512 gemm for Q4_0_8_8 (#9532)
|
hai 1 ano |
|
Georgi Gerganov
|
37f8c7b4c9
perplexity : remove extra new lines after chunks (#9596)
|
hai 1 ano |
|
Georgi Gerganov
|
bf9c1013ac
metal : use F32 prec for K*Q in vec FA (#9595)
|
hai 1 ano |
 Akarshan Biswas
|
e62e9789cd
Revert "[SYCL] fallback mmvq (#9088)" (#9579)
|
hai 1 ano |
|
R0CKSTAR
|
c35e586ea5
musa: enable building fat binaries, enable unified memory, and disable Flash Attention on QY1 (MTT S80) (#9526)
|
hai 1 ano |
 Molly Sophia
|
912c331d3d
Fix merge error in #9454 (#9589)
|
hai 1 ano |
 Johannes Gäßler
|
a5b57b08ce
CUDA: enable Gemma FA for HIP/Pascal (#9581)
|
hai 1 ano |
 Shankar
|
ecd5d6b65b
llama: remove redundant loop when constructing ubatch (#9574)
|
hai 1 ano |
|
Molly Sophia
|
2a63caaa69
RWKV v6: RWKV_WKV op CUDA implementation (#9454)
|
hai 1 ano |
 slaren
|
d09770cae7
ggml-alloc : fix list of allocated tensors with GGML_ALLOCATOR_DEBUG (#9573)
|
hai 1 ano |
