ngxson/llama.cpp

Pull Requests Commits

Apply suggestions from code review

ngxson committed 217 days ago

Verified 987955f0

rm double check for out-of-context

ngxson committed 219 days ago

3012326f

Merge branch 'master' into xsn/server_separate_pos_tokens

ngxson committed 219 days ago

3970042d

rm notion of n_kv_tokens

ngxson committed 219 days ago

de8956ad

parallel : add option for non-shared and larger prompts (#13598)

ggerganov committed 219 days ago

Verified 518329b2

vulkan: move common FA code to flash_attn_base.comp (#13556)

jeffbolznv committed 219 days ago

Verified 2f5a4e1e

vulkan: use scalar FA rather than coopmat2 when N==1 (#13554)

jeffbolznv committed 219 days ago

Verified 4f41ee11

llguidance : official v0.7.20 release (no actual changes) [noci] (#13594)

CoffeeVampir3 committed 219 days ago

Verified 3e0be1ca

server : do not return error out of context (with ctx shift disabled) (#13577)

ngxson committed 220 days ago

Verified 6aa892ec

webui : improve accessibility for visually impaired people (#13551)

ngxson committed 220 days ago

Verified aea9f8b4

readme : add list of dependencies and their license (#13591)

ngxson committed 220 days ago

Verified 06c1e4ab

releases : use arm version of curl for arm releases (#13592)

slaren committed 220 days ago

Verified 415e40a3

metal : add FA-vec kernel for head size 64 (#13583)

ggerganov committed 220 days ago

Verified 654a6779

llama : print hint when loading a model when no backends are loaded (#13589)

slaren committed 220 days ago

Verified 5364ae4b

fix stats report

ngxson committed 220 days ago

f9cc9f23

ci : add ppc64el to build-linux-cross (#13575)

CISC committed 220 days ago

Verified 7c07ac24

no more scan loop in n_kv_tokens()

ngxson committed 220 days ago

678d7b15

sycl : fixed compilation warnings (#13582)

lslusarczyk committed 220 days ago

Verified 0a338ed0

minja: sync (qwen3) (#13573)

ochafik committed 220 days ago

Verified bc098c3c

server : separate the notion of position and KV tokens, remove prompt truncation

ngxson committed 220 days ago

536bea4c

gguf : use ggml log system (#13571)

slaren committed 221 days ago

Verified c6a2c9e7

gguf-py : fix disconnect-before-connect in editor-gui (#13569)

danielzgtg committed 221 days ago

Verified 07ad2b6d

convert : fix conversion for llama 4 (#13567)

ngxson committed 221 days ago

Verified c531edfa

sycl: simplify bin_bcast_kernel (#13383)

AD2605 committed 221 days ago

Verified 02cdd2d8

sycl: reordered Q4_K MMVQ (#13109)

sgeor255 committed 221 days ago

Verified 64bb51cf

sycl: use oneDNN for matrices multiplication (#12972)

lslusarczyk committed 221 days ago

Verified 9c404ed5

llama-bench : fix -ot with dl backends (#13563)

slaren committed 221 days ago

Verified 6c8b9150

webui : handle PDF input (as text or image) + convert pasted long content to file (#13562)

ngxson committed 221 days ago

Verified 3cc1f1f1

server : proper error handling for missing elements in messages array (OpenAI compatible backend) (#13540)

pwilkin committed 221 days ago

Verified c753d7be

bench : handle decode errors (#13548)

ggerganov committed 221 days ago

Verified b2838049

Older