# microsoft/tensorwatch

Debugging, monitoring and visualization for Python Machine Learning and Data Science

Repository: https://github.com/microsoft/tensorwatch
Canonical: https://ross.abutalabs.com/products/tensorwatch
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: ai, deep-learning, deeplearning, machine-learning, machinelearning, reinforcement-learning, debugging, debugging-tool, debug, monitoring, explainable-ai, explainable-ml, saliency, model-visualization, data-science, python, jupyter-notebook, jupyter
Last push: 2026-03-30T21:27:28+00:00

## Health v2 (maintenance only)
Score: 66/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 74, release rhythm 35, longevity 100
- inputs: {"age_days": 2667, "days_push": 156, "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 3471, forks 361 (observed 2026-08-28T04:08:05.966389+00:00)

## What it is
TensorWatch is a Python library from Microsoft Research for debugging, monitoring, and visualizing machine learning training in real time, primarily within Jupyter Notebooks. It supports live queries against running training processes via its Lazy Logging mode and includes model/data analysis and visualization tools.

## Use cases
- visualize training metrics in real time in jupyter
- debug pytorch model training
- query live ml training process with arbitrary expressions
- explain model predictions with saliency maps
- plot neural network architecture diagrams
- monitor reinforcement learning experiments

## When to choose
- you train models in Jupyter Notebook and want live charts of metrics
- you need interactive, hackable debugging of PyTorch or TensorFlow eager training
- you want explainability tools like saliency visualizations for your models

## When to avoid
- production or multi-tenant environments - it can execute arbitrary code by design
- you need a polished, actively developed tool with broad framework support
- you only need simple metric tracking without interactive debugging

## Facets
- artifact type: library
- maturity: maintenance
- function: monitoring, data-visualization, logging, machine-learning, deep-learning
- domain: machine-learning, deep-learning, data-science, developer-tools
- platform: python, cross-platform
- tags: jupyter-notebook, debugging, explainable-ai, reinforcement-learning, pytorch, real-time-visualization, lazy-logging, jupyter

## Member repositories
- microsoft/tensorwatch (main) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:05.966389+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 2026-08-29T18:36:38.185231+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/tensorwatch (fetched 2026-08-28T04:08:05.966389+00:00, sha e38b9f2d03a2)
  - registry_pypi: https://pypi.org/pypi/tensorwatch/json (fetched 2026-08-29T09:30:35.607669+00:00, sha 6a4c353094e8)
- Data as of 2026-08-30T08:39:29.467469+00:00.