LLM inference in C/C++
C++ ggml
Commits on master
minor: fix reasoning preserve var for DS4 [no ci] (#25999)
e8e6c7a Piotr Wilkin (ilintar) committed 4 months ago
common: infer the speculative type from the draft repo sidecars (#25989)
6d5a910 Pascal committed 4 months ago
mtmd : use align_corners for qwen3vl vision position embedding interpolation (#25781)
b4aa7dd Gerben van V committed 4 months ago
hexagon: check tensor type when reusing descriptors (#25968)
71102a7 Wei Wang committed 4 months ago
kleidiai : warn once when a weight type has no KleidiAI kernel (#25701)
fb0e6b6 Kamalesh VS committed 4 months ago
server: return 400 instead of 500 on validation error with X-Conversation-Id (#25760)
fd41bf6 Pascal committed 4 months ago
vulkan: Refactor vk_queue to use per-instance mutexes and unique handles (#23570)
f048010 Winston Ma committed 4 months ago
ggml-openvino: Add GGML_BACKEND_DL_IMPL invocation for OpenVINO backend (#25795)
5735e10 Markus Ebner committed 4 months ago
CUDA: vectorize same-type get_rows with int4 copy (#25929)
305ba51 Piotr Wilkin (ilintar) committed 4 months ago
ui: Sidebar Conversations Bulk Action + Improved Settings logic/UI (#25815)
2beefef Aleksander Grygier committed 4 months ago
ui: fix collapsed user bubble with markdown rendering (#25869)
4ee6a9a Pascal committed 4 months ago
ui: enable the agentic flow when only the JS sandbox is active (#25865)
1521a9a Pascal committed 4 months ago
opencl: Support broadcast for Adreno MUL_MAT and honor `view_offs` for Adreno Q8_0 MUL_MAT for llama-server multi-stream (#25910)
178a6c4 Hongqiang Wang committed 4 months ago
llama-quant : exclude i32 ffn_gate_tid2eid routing table from quantization (#25787)
4937ca8 Yash Raj Pandey committed 4 months ago
opencl: load and use `kernel_gemm_moe_q6_k_f32_ns` from bin kernel lib (#25797)
86a9c79 lhez committed 4 months ago