ngxson/llama.cpp

Pull Requests Commits
llama : llm_type order by size (#13177)
CISC committed 345 days ago
Verified 7d3af70b
mtmd : add qwen2vl and qwen2.5vl (#13141)
ngxson committed 345 days ago
Verified 00e3e5a1
llama : set qwen3 model type sizes (#13175)
CISC committed 345 days ago
Verified e98b3692
llama-graph : fix text position for mrope (#13159)
ngxson committed 345 days ago
Verified b6ce7430
model : Nomic Embed Text V2 with Mixture-of-Experts (MoE) architecture (#12466)
manyoso committed 345 days ago
Verified 5f5e39e1
clip : fix model size display (#13153)
ngxson committed 345 days ago
Verified eaea3253
fix(rpc): Improve input validation and error handling (#13069)
thevilledev committed 345 days ago
Verified 43ddab6e
llama-bench: add `-d` depth arg (#13096)
thevishalagarwal committed 345 days ago
Verified 1831f538
mtmd : fix glm-edge redundant token count (#13139)
ngxson committed 345 days ago
Verified 4e87962e
context : do not clear output buffer on reserve (#13152)
pockers21 committed 345 days ago
Verified fb0471d1
llama : (mrope) allow using normal 1D position for text token (#13138)
ngxson committed 346 days ago
Verified d2b2031e
clip : refactor set input for cgraph + fix qwen2.5vl input (#13136)
ngxson committed 346 days ago
Verified 5fa9e63b
SYCL: Add all missing unary kernels (#13074)
qnixsynapse committed 346 days ago
Verified a4c340f9
readme : update hot topics (#13150)
ggerganov committed 346 days ago
Verified d0a417f3
common : fix noreturn compile warning (#13151)
ggerganov committed 346 days ago
Verified 43f2b071
llama-chat : fix typo GML --> GLM (#13143)
ngxson committed 346 days ago
Verified e5d6c255
musa: fix typo in cc control (#13144)
yeahdongcn committed 346 days ago
Verified f0dd6a19
CUDA: fix q_nope_absorbed prec for DS 2 Lite f16 (#13137)
JohannesGaessler committed 346 days ago
Verified 69699be4
arg : fix unused variable (#13142)
ngxson committed 346 days ago
Verified 85f36e5e
llama-bench : Add `--override-tensors` arg (#12922)
4onen committed 346 days ago
Verified c0a97b76
llama-chat : fix wrong template in GLM4-0414 (#13140)
matteoserva committed 346 days ago
Verified ced44be3
musa: fix build warning (#13129)
yeahdongcn committed 347 days ago
Verified e291450b
Fixes Qwen2.5VL segfault during inference with https://github.com/ggml-org/llama.cpp/pull/12402 as has_qwen2vl_merger migration was incomplete (#13133)
LostRuins committed 347 days ago
Verified 59e991c2
clip : Add Qwen2.5VL support (#12402)
HimariO committed 347 days ago
Verified ca2bb89e
common : add common_remote_get_content (#13123)
ngxson committed 347 days ago
Verified 2d451c80
clip : improve projector naming (#13118)
ngxson committed 347 days ago
Verified 4753791e
ggml: move fp16/bf16 conversion optimizations to CPU backend + export conversion APIs (#13107)
SongXiaoXi committed 347 days ago
Verified 77d5e9a7
grammar : handle maxItems == 0 in JSON schema (#13117)
rick-github committed 348 days ago
Verified d5fe4e81
llama : fix K-shift with quantized K and BLAS backend (#13113)
slaren committed 348 days ago
Verified 295354ea
Force FP32 compute in GLM4 FFN Down (#13101)
city96 committed 349 days ago
Verified 558a7647
Terms Of Use Privacy Policy Refund Policy Impressum