LLM inference in C/C++
C++ ggml
Commits on master
server: decode Windows OEM output to UTF-8 in built-in tools (#26597)
4308a4f Pascal committed 3 months ago
mtmd: correcting duplicate empty audio chunks for short inputs (#26536)
474c92e Abhinay Krishna committed 3 months ago
sampler : remove "full-context windows" from history-based samplers (#26524)
a6aa6f5 Oliver Simons committed 3 months ago
gguf-split: Add option to delete split parts during merge (#26538)
76c956c Guilherme Quintino committed 3 months ago
mtmd: support Qwen3-TTS (note: breaking change to llama-tts binary) (#26254)
0713275 Xuan-Son Nguyen committed 3 months ago
ci: fix pre-built binaries no longer working on macOS 15 and below (#26375)
6b5224c Niklas Wenzel committed 3 months ago
speculative : refactor enabled configs common_speculative_init (#26510)
7bd8282 Daniel Bevenius committed 3 months ago
gguf-py: validate n_dims and guard against uint64 overflow in reader (#25401)
5788b51 hcl committed 3 months ago
convert : add missing return after setting tekken vocab (#25947)
b5746d2 Angel Galindo committed 3 months ago
vulkan backend ops: implemented GATED_LINEAR_ATTN (#25601)
f26efa0 Pranav Uttarkar committed 3 months ago
convert : import bytes_to_unicode from convert_slow_tokenizer (#26217)
b06fbc9 Caleb DeLeeuw committed 3 months ago
model : allow reshape of tensors during load (#26531)
1269cb1 Georgi Gerganov committed 3 months ago
llama : move n_vocab from llama_sampler_data to penalty_sampler (#26520)
935cad6 Oliver Simons committed 3 months ago
sycl: parallelize the non-contiguous concat kernel (#25852)
6c8dcaa Titaniumtown committed 3 months ago
Extended SYCL oneDNN SDPA to non-FP16 KV caches (Q4_0–Q8_0 and FP32) (#25874)
66fa168 Ozymandias_EBON committed 3 months ago
chat : add new template for DeepSeek V4 Flash 0731 (#26398)
0ef6e55 Thiago Padilha committed 3 months ago
vendor : update cpp-httplib to 0.52.0 (#26485)
94bc47f Alessandro de Oliveira Faria (A.K.A.CABELO) committed 3 months ago
vendor : update BoringSSL to 0.20260803.0 (#26523)
fe2adf0 Alessandro de Oliveira Faria (A.K.A.CABELO) committed 3 months ago
tests: add model resolution test on synthetic repo listings (#26172)
ee0445c Pascal committed 3 months ago
ggml: use dynamic allocation for split graph inputs (#22789)
dbadb68 AgoraPete committed 3 months ago
opencl: route large q6_K lm_head to the flat GEMV (#26427)
39eab74 Hongqiang Wang committed 3 months ago