ggerganov/llama.cpp

Pull Requests Commits

server: keep a queued model out of the victim pool until its waiters leave

ServeurpersoCom committed 68 days ago

eee2fa9c

server: fix LRU hang on multiple requests same model

ngxson committed 68 days ago

f2dc72b8

caps : recheck typed content if template checks for string (#28511)

CISC committed 68 days ago

Verified 5202104b

convert : write explicit recurrent_layers for Qwen3-Next / Qwen3.5 (#28208)

devvexus committed 68 days ago

Verified 9a757058

ggml-cuda: fix divergent barrier in f16 flash attention (#27870)

siavashnorouzi committed 68 days ago

Verified b74f590e

ggml: allow backend inputs to not create another split (#28387)

am17an committed 68 days ago

Verified 992cb503

vulkan: rms_norm fusion opportunities (#28024)

jeffbolznv committed 68 days ago

Verified 9ac8c408

ci : add container image checking and tagging (wip) (#28394)

danbev committed 68 days ago

Verified 2092353c

vulkan: add TQ1_0 support (mm, mat-vec, mat-vec-id, dequant, get_rows) (#27765)

Anjielon committed 68 days ago

Verified 8fe90e1f

convert : add `--fuse-qkv` flag to fuse Q/K/V into QKV during HF-to-GGUF conversion (#22780)

JoursBleu committed 69 days ago

Verified 465e49b9

models : fix GDN normalization from `max` to `rsqrt` (#28068)

danielhanchen committed 69 days ago

Verified 5fdfa628

[Model] Support for Spark2_5ForCausalLM implementation (#27868)

KnightYao committed 69 days ago

Verified 3ad1ba73

opencl: properly choose weights pack for q4_K, q5_K mul_mat (#28402)

lhez committed 69 days ago

Verified d03efa5d

cuda: fixes races in mmid and mmf (#28475)

am17an committed 69 days ago

Verified 73a43d1f

grammar : fix max repetition threshold (#28469)

aldehir committed 69 days ago

Verified 9e0e2205

ui: Improve Chat Messages rendering performance (#28460)

allozaur committed 69 days ago

Verified 0afb805b

common: add --log-jsonl (#28437)

ngxson committed 69 days ago

Verified 7620399f

ui : embed assets directly with CMake (#28445)

angt committed 69 days ago

Verified c457e3bf

metal : add remaining fa-vec tunings for M2 Max (#28458)

nikwen committed 69 days ago

Verified 971595d6

Github: limit blank issues to maintainers (#28435)

JohannesGaessler committed 70 days ago

Verified 74a7c897

metal : fix memory leak in early return (#28399)

nikwen committed 70 days ago

Verified 6a1a922d

sycl : fix test-backend-ops CI break && restore Kronecker product FWHT support (#28016) (#28254)

philip-jingxin committed 70 days ago

Verified 4d917609

sycl: attribute device allocations by site (GGML_SYCL_MEMTRACE) (#27631)

nicois committed 70 days ago

Verified cd8cdf39

metal : add remaining fa-vec tunings for M3 (#28396)

init-22 committed 71 days ago

Verified 427291b5

ui : fix MCP image attachments not displayed in tool block (#25789) (#28089)

nachobh committed 71 days ago

Verified 85d5703a

opencl: extend the elementwise and data‐movement op coverage (#27633)

wanghqc committed 71 days ago

Verified 1548a240

opencl: add Adreno xmem SDPA path (#26331)

happyyzy committed 71 days ago

Verified 4acf4a4c

ci : move more jobs to ccache-buckets (#28375)

CISC committed 71 days ago

Verified 8b4b3558

ui: export conversations from database instead of cached store (#27432)

intel00000 committed 71 days ago

Verified 1863ac03

model : add Tencent Hy 4 (hy_v4) preview architecture support (#28127)

Little0o0 committed 71 days ago

Verified 49c0dc82

Older