LLM inference in C/C++
C++ ggml
Commits on master
revert to using global_invocation_id for cpy shader (#23955)
b8275a8 Masashi Yoshimura committed 5 months ago
speculative : fix n_outputs_max and remove draft-simple auto-enable (#23988)
5dcb711 Georgi Gerganov committed 5 months ago
nix : add nix-nodejs facilities to build Web UI (#23846)
5aa3a64 Christian Hoener zu Siederdissen committed 5 months ago
metal: template GLU kernels to support f16/f32 (#23882)
95b8b8e Shrivas Shankar committed 5 months ago
vulkan: don't hold the device mutex while compiling pipelines (#23641)
55ac090 Jeff Bolz committed 5 months ago
vulkan: Block-load Q3_K/Q6_K block data and subtract on 32b ints (#23056)
1962000 Matt Corallo committed 5 months ago
common : support manually triggering the reasoning budget end sequence (#23949)
5254a79 Aldehir Rojas committed 5 months ago
ci : add missing Linux label to cpu-x64-high-perf runner (#23958)
e22b0de Georgi Gerganov committed 5 months ago
[SYCL] Support Q4_1, Q5_0, Q5_1 in Flash-attention (#23812)
a511424 Neo Zhang committed 5 months ago
ci : limit trigger paths for the CPU workflow (#23938)
399739d Georgi Gerganov committed 5 months ago
vocab : add tokenizer support for jina-embeddings-v2-base-zh (#18756)
d4c8e2c o7si committed 5 months ago