# haitongli/knowledge-distillation-pytorch

A PyTorch implementation for exploring deep and shallow knowledge distillation (KD) experiments with flexibility

Repository: https://github.com/haitongli/knowledge-distillation-pytorch
Canonical: https://ross.abutalabs.com/products/knowledge-distillation-pytorch
Language: Python
License: MIT
License Family: permissive
Topics: pytorch, knowledge-distillation, deep-neural-networks, cifar10, model-compression, dark-knowledge, computer-vision
Last push: 2023-03-25T00:03:08+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": 3099, "days_push": 1258, "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 2000, forks 353 (observed 2026-08-28T04:06:03.846543+00:00)

## What it is
A PyTorch framework for running knowledge distillation experiments, supporting both shallow (teacher-to-small-CNN) and deep distillation on CIFAR-10. It includes pretrained teacher checkpoints, hyperparameter search, and result synthesis utilities.

## Use cases
- distill a large model's knowledge into a smaller network
- run knowledge distillation experiments in pytorch
- compress a resnet into a lightweight cnn for cifar-10
- search hyperparameters for distillation training
- learn how knowledge distillation works with example code
- train student models using dark knowledge from pretrained teachers

## When to choose
- you want a ready-to-run PyTorch codebase for KD experiments on CIFAR-10
- you need pretrained teacher models and hyperparameter search tooling for distillation
- you are exploring model compression for efficient hardware deployment

## When to avoid
- you need distillation for datasets other than CIFAR-10 or for transformer/LLM models
- you want a production-grade, actively maintained training framework
- you need a plug-and-play library API rather than a research codebase driven by CLI scripts

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, llm-training
- domain: deep-learning, computer-vision, machine-learning
- platform: python
- tags: knowledge-distillation, model-compression, pytorch, cifar10, dark-knowledge, teacher-student, research-code, linux, macos, gpu

## Member repositories
- haitongli/knowledge-distillation-pytorch (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:03.846543+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 2026-08-30T03:02:07.749489+00:00, confidence not recorded.
  - readme: https://github.com/haitongli/knowledge-distillation-pytorch (fetched 2026-08-28T04:06:03.846543+00:00, sha 0bc77c80029c)
- Data as of 2026-08-30T08:39:29.467469+00:00.