78 const std::string& tools_json);
91 const std::string& raw);
ChatAdapter concrete base class.
Concrete base class for chat format adapters (80% logic).
LlamaCppBackend — common llama.cpp patterns (15% layer).
Activate model on GPU (WARM → ACTIVE).
bool calls_satisfy_schema(const std::vector< ToolCall > &calls, const std::string &tools_json)
Check that every call's declared required parameters are present.
ParsedModelResponse parse_model_response(LlamaCppBackend *llama, ChatAdapter *adapter, const std::string &raw)
Parse a raw emission: template first, adapter second.
Content and tool calls extracted from one raw emission.
bool used_template
True when common_chat produced the calls.
std::vector< ToolCall > tool_calls
Extracted calls.
std::string content
Reasoning-stripped content.