AnalysisDevelopers
23 hours ago
llama.cpp adds support for Step3.5/3.7 flash mtp3
Pull request #24340 adds multi-layer MTP support for Step3.5/3.7 models in llama.cpp. It is a follow-up to PR #23274.
Pull request #24340 adds multi-layer MTP support for Step3.5/3.7 models in llama.cpp. It is a follow-up to PR #23274.