feat(backend): add buun-llama-cpp fork (DFlash + TCQ KV-cache)#9532

Open

mudler wants to merge 8 commits into

feat/buun-llama-cpp-backend

Commits on Apr 24, 2026

feat(backend): add buun-llama-cpp fork (DFlash + TCQ KV-cache)
mudler
committed
test(gallery): extend importer specs to cover buun-llama-cpp
mudler
committed
fix(buun-llama-cpp): backport logit_bias_eog field to grpc-server copy
mudler
committed
fix(buun-llama-cpp): drop logit_bias_eog arg from params_from_json_cmpl
mudler
committed
ci(buun-llama-cpp): wire backend into test-extra + build matrix
mudler
committed
fix(buun-llama-cpp): shim atomicAdd(double*,double) for pre-sm_60 CUDA
mudler
committed
fix(buun-llama-cpp): pass WARP_SIZE to argmax __shfl_xor_sync calls
mudler
committed
fix(buun-llama-cpp): shim cudaMemcpy{To,From}Symbol + WARP_SIZE on fwht128 shuffles
mudler
committed