PR #19802 llama: end-to-end tests

tests: add end-to-end tests per model architecture

JohannesGaessler committed 242 days ago

fixup for rebase

JohannesGaessler committed 242 days ago

fix use-after-free in llama-model-loader.cpp

JohannesGaessler committed 242 days ago

fix CI

JohannesGaessler committed 242 days ago

fix WebGPU

JohannesGaessler committed 242 days ago

fix CI

JohannesGaessler committed 242 days ago

disable CI for macOS-latest-cmake-arm64

JohannesGaessler committed 242 days ago

use expert_weights_scale only if != 0.0f

JohannesGaessler committed 241 days ago

comments [no ci]

JohannesGaessler committed 240 days ago

llama.cpp llama: end-to-end tests #19802 Merged