ggml-org/llama.cpp

Pull Requests Commits

ui : put the discover view in the models manager

allozaur committed 5 hours ago

694f2384

ui : add the discover models components

allozaur committed 5 hours ago

3f56d5b3

ui : add the models manager

allozaur committed 5 hours ago

Verified 6226aae1

ui : add the shared model row and list primitives

allozaur committed 5 hours ago

Verified bd03fbcb

ui : render shared model row hints as native titles

allozaur committed 5 hours ago

Verified 31220d5c

ui : remember hub avatars that failed to load

allozaur committed 5 hours ago

Verified d5219811

ui : shared model display primitives

allozaur committed 5 hours ago

Verified bf540ab0

ui : optional sanitized raw HTML in markdown

allozaur committed 5 hours ago

Verified d4e86b6e

ui : model download pipeline

allozaur committed 5 hours ago

Verified 509938e8

ui : model memory-fit estimation

allozaur committed 5 hours ago

Verified 5b89ec38

ui : strip provider tilde prefix from hub avatar urls

allozaur committed 5 hours ago

Verified 824f42c3

ui : Hugging Face Hub data layer

allozaur committed 5 hours ago

Verified ca2d8565

ui : model id grammar for sidecars, quants and capability parsing

allozaur committed 5 hours ago

Verified 762e3393

ui : type-safe API types, fetch helpers and download-ready models store plumbing

allozaur committed 5 hours ago

Verified e54a71f4

common : fix HF cache paths on Windows (#29475)

angt committed 6 hours ago

Verified 6c7a87f7

webgpu: Handle unaligned writes in ggml_backend_webgpu_buffer_set_tensor (#29471)

jbooth committed 6 hours ago

Verified f00a64c1

tests : refactor test-recurrent-state-rollback (#29426)

ggerganov committed 6 hours ago

Verified d77dd080

ggml-cpu: enable tiled flash attention for non-vector-multiple head dims on x86 (#29423)

SongXiaoXi committed 7 hours ago

Verified 6f767fe9

ci : ignore more vgpr spills in > 256 DQK fattn kernels (#29571)

IMbackK committed 7 hours ago

Verified f916130d

vulkan: fuse qwen4exp's SCALE -> SIGMOID -> SCALE -> hc_post chain (#29520)

fxgsell committed 8 hours ago

Verified 03a667aa

HIP: fix template skip for DKQ > 256 mfma kernels (#29559)

IMbackK committed 8 hours ago

Verified c2a9e160

metal: support left and circular padding in GGML_OP_PAD (#29561)

ServeurpersoCom committed 10 hours ago

Verified 4364bf72

context : do not re-reserve the scheduler when toggling causal_attn (#28751)

sihanyu03 committed 11 hours ago

Verified ed7ac35e

Enables Windows ARM64 build with MSVC cl.exe (#28362)

sarahwu185 committed 12 hours ago

Verified 0c6a6a7c

tests : fix ggml init (#29554)

ggerganov committed 13 hours ago

Verified 81ef10ea

vulkan: fix wrong results when a mul_mat reads a slice of a larger cache (#28956)

ServeurpersoCom committed 13 hours ago

Verified 52624716

server : allow RANK pooling batch splitting for causal LLM rerankers (ie. Qwen3 and Qwen3-VL) (#28876)

timothywang21 committed 23 hours ago

Verified 4da63377

common : avoid side effects around params parsing (#29537)

ggerganov committed 1 day ago

Verified a97cce86

common : make string_split<T> throw on invalid values (#29518)

angt committed 1 day ago

Verified 136887b6

convert : export YaRN scaling parameters for PLaMo-3 (#29528)

tokinasin committed 1 day ago

Verified 9adc7f42

Older