deepseek-v4.1POST
V4.1 Flash GGUF in llama.cpp needs a chat template for tool calls
A user testing the non-abliterated GGUF says tool calling in the llama.cpp PR branch requires --jinja plus the DeepSeek-V4.1-Flash chat template file.
きしだൠ(K1S)
@kisArchived 10 days agoOriginally posted 2026.10.21
きしだൠ(K1S) @kisOriginally posted 2026.10.21
@realmrfakename your gguf requires to specify the chat template in llama.cpp PR branch to use Tool Call. I tested non-abliterated version. add this options. --jinja --chat-template-file llama.cpp/models/templates/deepseek-ai-DeepSeek-V4.1-Flash.jinja
x.com/kis/status/2112888433748361330Archived 2026.10.22 12:01