ngxson/llama.cpp

Pull Requests Commits

add clarify comment for norm eps

ngxson committed 229 days ago

Verified 1c0a25ba

restore boi/eoi check

ngxson committed 229 days ago

321d4ed1

add to list of pre-quant

ngxson committed 229 days ago

d0adae42

add test for internvl

ngxson committed 229 days ago

95be3888

ngxson committed 229 days ago

abb12389

rm mobilevlm from test

ngxson committed 229 days ago

6472d8c8

ngxson committed 229 days ago

ae25cd29

InternVL3-1B working

ngxson committed 229 days ago

2bfa7a4c

convert : internvl support

ngxson committed 229 days ago

e309f161

CUDA: fix FlashAttention on Turing (#13415)

JohannesGaessler committed 229 days ago

Verified d8919424

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

ngxson committed 229 days ago

Verified 7fef1176

vulkan: scalar flash attention implementation (#13324)

jeffbolznv committed 229 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 230 days ago

Verified 33eff402

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

Alberto Cabrera Pérez committed 230 days ago

Verified 17512a94

metal : optimize MoE for large batches (#13388)

ggerganov committed 230 days ago

Verified 611aa914

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

JohannesGaessler committed 230 days ago

Verified 0cf6725e

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

slaren committed 230 days ago

Verified 27ebfcac

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

JohannesGaessler committed 230 days ago

Verified 5c86c9ed

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

bartowski1182 committed 230 days ago

Verified efb8b47e

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

yeahdongcn committed 230 days ago

Verified 0527771d

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

ngxson committed 230 days ago

Verified 2189fd3b

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

ngxson committed 230 days ago

Verified 3f96aeff

rpc : add rpc_msg_set_tensor_hash_req (#13353)

rgerganov committed 230 days ago

Verified b486ba05

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

jeffbolznv committed 230 days ago

Verified 02115dcd

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

ngxson committed 230 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 231 days ago

Verified ee01d71e

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

ngxson committed 231 days ago

Verified 8c83449c

Older