Implementation:Ollama Ollama Llama Model InternLM2

Knowledge Sources	Ollama
Domains	LLM Inference, Model Architecture
Last Updated	2025-02-15 00:00 GMT

Overview

Implements the ggml computation graph builder for the InternLM2 model architecture.

Description

The llm_build_internlm2 constructor builds a standard transformer with RoPE-based positional encoding, RMS-normalized self-attention with Q/K/V projections and optional biases, and SiLU-gated feed-forward layers across all transformer blocks. The architecture follows a standard dense transformer pattern with grouped-query attention support.

Usage

Enables Ollama to run InternLM2 family models through the llama.cpp inference engine.

Code Reference

Source Location

• Repository: Ollama
• File: llama/llama.cpp/src/models/internlm2.cpp
• Lines: 1-120

Signature

llm_build_internlm2::llm_build_internlm2(
    const llama_model & model,
    const llm_graph_params & params) : llm_graph_context(params);

Import

#include "models.h"

I/O Contract

Inputs

Name	Type	Required	Description
model	const llama_model &	Yes	Loaded model with InternLM2 weights
params	const llm_graph_params &	Yes	Graph construction parameters

Outputs

Name	Type	Description
ggml graph	ggml_cgraph	Complete InternLM2 computation graph

Usage Examples

auto builder = llm_build_internlm2(model, params);

Related Pages

• Principle:Ollama_Ollama_LLM_Inference_Pipeline