nemotron3_adapter.cpp File Reference

Nemotron 3 adapter implementation (v2.1.9, gh#47; v2.3.8, gh#70). More...

#include "nemotron3_adapter.h"
#include <entropic/types/logging.h>
#include <nlohmann/json.hpp>
#include <regex>
#include <sstream>

Include dependency graph for nemotron3_adapter.cpp:

Go to the source code of this file.

Namespaces
namespace	entropic
	Activate model on GPU (WARM → ACTIVE).

Functions
static nlohmann::json	entropic::build_tool_defs (const std::vector< std::string > &tool_jsons)
	Build the OpenAI-function <tools> JSON array for injection.

Detailed Description

Nemotron 3 adapter implementation (v2.1.9, gh#47; v2.3.8, gh#70).

Tool-call parsing targets the DSML invoke format the nemotron_h GGUFs actually emit (gh#70), with the qwen3_coder XML and tagged-JSON paths retained as backstops. The chat template itself is GGUF-embedded so chat_format() returns an empty string and llama.cpp drives the template application.

Version

2.3.8

Definition in file nemotron3_adapter.cpp.