Skip to content

LLM Compiler-7b-ftd

A Code Llama model for compiler optimisation tasks

Visit official site

What LLM Compiler-7b-ftd is

LLM Compiler-7b-ftd is described as a Meta model based on Code Llama, adapted for code optimisation and compiler reasoning tasks around LLVM assembly.

What you can do with it

  • Predict optimisation effects
  • Generate optimisation sequences
  • Disassemble assembly into LLVM IR

Application scenarios

  • Shrink a compiled program
  • Tune an optimisation sequence
  • Prepare code for further analysis

Getting started

  • Install the required libraries
  • Load the tokenizer from the pretrained model
  • Build a text generation pipeline
  • Feed in the code fragment to optimise

Pros & cons

✓ What we like

  • A text generation pipeline is the documented entry point
  • The model is aimed at compiler research and product work

! What to watch out for

  • Loading needs the transformers library
  • Code fragments are passed to a pipeline as input

FAQ

What is it based on?

The material states it is built on Code Llama and developed by Meta.

What does it predict?

Optimisation effects on LLVM assembly code are listed.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review