ggml-org/llama.cpp

Pull Requests Commits

ui : mark llama-app-reusable code with a LLAMA-APP-REUSE tag

allozaur committed 2 days ago

5c648695

ui : fix double blank line in model types

allozaur committed 2 days ago

5b7a4340

ui : add download state tracking to the model status manager

allozaur committed 2 days ago

14ac965b

ui : add model download pipeline

allozaur committed 2 days ago

0d22938c

ui : simplify model memory estimation

allozaur committed 2 days ago

a07359ef

ui : add model compatibility estimation

allozaur committed 2 days ago

af98525e

ui : move catalog url to models-discover constants

allozaur committed 2 days ago

548acc2f

ui : move HF constants to constants and enums modules

allozaur committed 2 days ago

7e63e9e3

ui : follow MODEL_ID regex rename in huggingface service

allozaur committed 2 days ago

ddb8d219

ui : add huggingface hub data layer

allozaur committed 2 days ago

4aa7063b

ui : lowercase reasoning capability and use sidecar enums in tests

allozaur committed 2 days ago

f6f90463

ui : address review on model id parsing

allozaur committed 2 days ago

4c93cef4

ui : parse sidecar types in model ids

allozaur committed 2 days ago

c1cbd5e2

ui : add format:files script for formatting changed files only

allozaur committed 2 days ago

4b5449a4

ui : rename model API types to mirror endpoints

allozaur committed 2 days ago

9a851978

ui : remove dead ChatFormActionAddMcpSubmenu component

allozaur committed 2 days ago

2853c181

metal : add fa-vec tunings for M1 Ultra (#28088)

ozgursoy committed 2 days ago

Verified 458681e1

CUDA: XOR swizzle flash attn K,V smem fp16 tiles (#25635)

ynankani committed 2 days ago

Verified e4b9af00

metal : add concat support for quantized types (#28116)

ggerganov committed 2 days ago

Verified ab0b3bd3

AVX2: Speed up large batch size prompt processing of IQ models (#27402)

bartowski1182 committed 2 days ago

Verified 85c55223

metal : add top-k radix implementation (#28073)

ggerganov committed 2 days ago

Verified 2a74817f

kv-cache : optimize restoring non-contiguous cells (#27991)

itsnotoger committed 2 days ago

Verified 2d8d612e

opencl: tune the quant paths for Intel Xe-LP GPUs to improve its TG and PP performance (#26438)

wanghqc committed 2 days ago

Verified 010be968

ui: copy the displayed text of grouped agentic responses (#27832)

ServeurpersoCom committed 2 days ago

Verified 774ee0e2

webgpu : avoid crash when offset is not multiple of 4 in WebGPU ggml_backend_tensor_get() implementation (#28045)

fairydreaming committed 3 days ago

Verified 8e53fcef

ROCm: add radix TOP_K for long rows (#27466)

jadenmach2 committed 3 days ago

Verified f8dbcd61

metal : add fa-vec tunings for M1 (#28078)

nikwen committed 3 days ago

Verified 5d4a3be2

CUDA: extend MOE fusion to specdec, earlier MOE glu fusion and topk-router fusion were restricted to 1 token (#27621)

ynankani committed 3 days ago

Verified 41ef91f7

sycl : Enhance to get the free memory of Intel GPU (#27968)

arthw committed 3 days ago

Verified a32af33d

ci : add check for unzip (#28082)

CISC committed 3 days ago

Verified 580e88d8

Older