ggml-org/llama.cpp

Pull Requests Commits

use scalar sums

0cc4m committed 27 days ago

715ed286

apply scales inline

0cc4m committed 28 days ago

a9435151

vulkan: add int8 coopmat quantized matmul shader

0cc4m committed 28 days ago

d1f8bbd0

opencl: add l2_norm (#20160)

lhez committed 28 days ago

Verified 6fce5c6a

Autoparser: True streaming (#20177)

pwilkin committed 28 days ago

Verified c024d859

Autoparser: add optional argument reshuffle capability (#20171)

pwilkin committed 28 days ago

Verified 2f2923f8

quants : Add memsets and other fixes for IQ quants (#19861)

bartowski1182 committed 28 days ago

Verified 649f0648

Add @pwilkin to CODEOWNERS for autoparser code (#20174)

pwilkin committed 29 days ago

Verified 74636871

Autoparser - complete refactoring of parser architecture (#18675)

pwilkin committed 29 days ago

Verified 566059a2

hexagon: add f32 ssm_conv op (#20122)

tboinovski1 committed 29 days ago

Verified 34df42f7

server : preserve anthropic thinking blocks in conversion (#20120)

T0mSIlver committed 29 days ago

Verified e68f2fb8

cpu: skip redudant ROPE cache updates (#20149)

max-krasnyansky committed 29 days ago

Verified ba2fd11c

ggml-cuda: add mem check for fusion (#19916)

am17an committed 29 days ago

Verified d48e8764

ggml: update comments for backends which have no memory to report (#20157)

taronaeo committed 29 days ago

Verified ba2ff79e

ggml-cpu: Fix gcc 15 ICE on ppc64le (#20083) (#20130)

shalinib-ibm committed 29 days ago

Verified c6980ff2

CUDA: use shared mem for ssm_conv (#20128)

am17an committed 29 days ago

Verified 1e38a7a6

context: ignore zero scale LoRAs when checking sameness (#20166)

TimNN committed 29 days ago

Verified 388baabc

Checkpoint every n tokens: squash (#20087)

pwilkin committed 29 days ago

Verified f5ddcd16

webui: Agentic Loop + MCP Client with support for Tools, Resources and Prompts (#18655)

allozaur committed 29 days ago

Verified f6235a41

ggml-cpu: fix data race for debug asserts (#20148)

JohannesGaessler committed 29 days ago

Verified 2850bc6a

kv-cache : fix M-RoPE checkpoints (#20132)

ggerganov committed 29 days ago

Verified 17a42589

cli : Don't clear system prompt when using '/clear' (#20067)

roj234 committed 29 days ago

Verified f7db3f37

opencl: add neg, exp and diag (#20127)

lhez committed 29 days ago

Verified 6c97bffd

hexagon: add fp16 support for binary ops: add,sub,mul,div (#20139)

YardenTal44 committed 29 days ago

Verified 2b10b626

models : kda chunk size = 16 (#19827)

ymcki committed 30 days ago

Verified a0ed91a4

CUDA: Improve performance via less synchronizations between token (#17795)

aendk committed 30 days ago

Verified 2cd20b72

model : update Qwen3.5 model type detection (#20126)

EZForever committed 30 days ago

Verified 872646b3

cli : add command and file auto-completion (#19985)

CISC committed 30 days ago

Verified b5ed0e05

convert : register Qwen 3.5 ForCausalLM for text only (#20119)

CISC committed 30 days ago

Verified cf232515

webui: Improvements for Models Selector UI (#20066)

allozaur committed 30 days ago

Verified 5e335ba1

Older