Implementation:Ollama Ollama Llama Model Grok

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 Grok model architecture from xAI.

Description

The llm_build_grok constructor builds a transformer with RoPE-based positional encoding, RMS-normalized self-attention with Q/K/V projections and optional biases, and Mixture-of-Experts feed-forward layers across all transformer blocks. Produces final logits via an output projection for autoregressive generation.

Usage

Enables Ollama to run xAI Grok models through the llama.cpp inference engine by defining how the model's MoE architecture maps to ggml tensor operations.

Code Reference

Source Location

• Repository: Ollama
• File: llama/llama.cpp/src/models/grok.cpp
• Lines: 1-159

Signature

llm_build_grok::llm_build_grok(
    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 Grok MoE weights
params	const llm_graph_params &	Yes	Graph construction parameters

Outputs

Name	Type	Description
ggml graph	ggml_cgraph	Complete Grok MoE computation graph

Usage Examples

auto builder = llm_build_grok(model, params);

Related Pages

• Principle:Ollama_Ollama_LLM_Inference_Pipeline