# VHellendoorn/Code-LMs

Guide to using pre-trained large language models of source code

Repository: https://github.com/VHellendoorn/Code-LMs
Canonical: https://ross.abutalabs.com/products/code-lms
Language: Python
License: MIT
License Family: permissive
Topics: gpt-2, deep-learning, source-code
Last push: 2024-07-07T01:08:30+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1742, "days_push": 788, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1842, forks 261 (observed 2026-08-28T04:05:43.162515+00:00)

## What it is
A guide to using pre-trained large language models of source code, centered on the PolyCoder models with instructions for Hugging Face inference, GPT NeoX training, and evaluation. It includes setup instructions, model checkpoints, datasets, and benchmarking guidance.

## Use cases
- use polycoder for code generation
- run a pretrained code language model locally
- evaluate code LLMs on humaneval
- learn how to use large models of source code
- load polycoder checkpoints with huggingface transformers
- replicate code model perplexity evaluation

## When to choose
- you want a practical guide to running PolyCoder or similar code LLMs
- you need instructions for GPT NeoX-based code model training and evaluation
- you want links to pretrained checkpoints and datasets for code models

## When to avoid
- you need a maintained production code-generation service
- you want a general-purpose LLM not specialized for source code
- you need up-to-date model coverage beyond 2022-era models

## Facets
- artifact type: learning-resource
- maturity: maintenance
- function: llm-training, llm-inference, machine-learning, documentation
- domain: large-language-models, deep-learning, developer-tools, tutorials
- platform: python
- tags: code-generation, polycoder, gpt-neox, huggingface, source-code-models, humaneval, gpu, docker, linux

## Member repositories
- VHellendoorn/Code-LMs (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:43.162515+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:17:58.862889+00:00, confidence not recorded.
  - readme: https://github.com/VHellendoorn/Code-LMs (fetched 2026-08-28T04:05:43.162515+00:00, sha 70781e554607)
- Data as of 2026-08-30T08:39:29.467469+00:00.