ngxson/llama.cpp

Pull Requests Commits

rm redundant include

ngxson committed 229 days ago

f691bb5e

fix windows build

ngxson committed 229 days ago

e48d52ca

mtmd : move helpers to dedicated file

ngxson committed 229 days ago

0ce5415c

llguidance : set tokenizer slices to default (#13424)

CISC committed 230 days ago

Verified 43dfd741

ci: free_disk_space flag enabled for intel variant (#13426)

Thammachart committed 230 days ago

Verified b064a51a

mtmd : support InternVL 2.5 and 3 (#13422)

ngxson committed 230 days ago

Verified 053367d1

CUDA: fix FlashAttention on Turing (#13415)

JohannesGaessler committed 230 days ago

Verified d8919424

arg : add env var to control mmproj (#13416)

ngxson committed 230 days ago

Verified 7fef1176

vulkan: scalar flash attention implementation (#13324)

jeffbolznv committed 230 days ago

Verified dc1d2adf

chore(llguidance): use tagged version that does not break the build (#13413)

HRKings committed 230 days ago

Verified 7c28a74e

server : vision support via libmtmd (#12898)

ngxson committed 231 days ago

Verified 33eff402

sycl : implementation of reordered Q4_0 MMVQ for Intel GPUs (#12858)

Alberto Cabrera Pérez committed 231 days ago

Verified 17512a94

metal : optimize MoE for large batches (#13388)

ggerganov committed 231 days ago

Verified 611aa914

CUDA: FA support for Deepseek (Ampere or newer) (#13306)

JohannesGaessler committed 231 days ago

Verified 0cf6725e

llama : do not crash if there is no CPU backend (#13395)

slaren committed 231 days ago

Verified 27ebfcac

CUDA: fix crash on large batch size for MoE models (#13384)

JohannesGaessler committed 231 days ago

Verified 5c86c9ed

imatrix : Add --parse-special for enabling parsing of special tokens in imatrix calculation (#13389)

bartowski1182 committed 231 days ago

Verified efb8b47e

llama-run: add support for downloading models from ModelScope (#13370)

yeahdongcn committed 231 days ago

Verified 0527771d

mtmd : fix batch_view for m-rope (#13397)

ngxson committed 231 days ago

Verified 2189fd3b

llama : one-off chat template fix for Mistral-Small-2503 (#13398)

ngxson committed 231 days ago

Verified 3f96aeff

rpc : add rpc_msg_set_tensor_hash_req (#13353)

rgerganov committed 231 days ago

Verified b486ba05

vulkan: Allow up to 4096 elements for mul_mat_id row_ids (#13326)

jeffbolznv committed 231 days ago

Verified 02115dcd

server : (webui) rename has_multimodal --> modalities (#13393)

ngxson committed 231 days ago

Verified d9c4acca

ci : limit write permission to only the release step + fixes (#13392)

slaren committed 231 days ago

Verified 15e03282

mtmd : Expose helper_decode_image_chunk (#13366)

mattjcly committed 231 days ago

Verified f05a6d71

server : (webui) fix a very small misalignment (#13387)

ngxson committed 232 days ago

Verified ee01d71e

server : (webui) revamp the input area, plus many small UI improvements (#13365)

ngxson committed 232 days ago

Verified 8c83449c

convert : support rope_scaling type and rope_type (#13349)

CISC committed 232 days ago

Verified 1a844be1

mtmd : fix the calculation of n_tokens for smolvlm (#13381)

awkrail committed 232 days ago

Verified 0ccc1213

context : allow cache-less context for embeddings (#13108)

ggerganov committed 232 days ago

Verified 6562e5a4

Older