MCP servers · aparajeets

TrainTools: paper-based training diagnostics for PyTorch

A set of small diagnostic tools for PyTorch and Hugging Face training runs, each answering one question about batch size, plasticity, gradients or example quality.

This text was machine-translated from the publisher's own information and has not been checked by a human editor. The publisher's original text is the reference.

This resource was imported automatically from the sources listed below and has not been reviewed by the AI Gate team yet. The description is the publisher's own, as published. Check the repository before installing.

What it does

TrainTools provides small diagnostic tools that answer practical questions during a training run. They cover whether batch size wastes compute (Gradient Noise Scale), whether a network is losing its ability to learn (PlasticityProbe), whether to stop training (TrainGuard), and whether gradients or batches are unhealthy. Example trackers flag forgotten, ambiguous or likely mislabeled examples. The methods follow published papers, which the documentation cites.

Use it while a run is in progress, when a loss curve or batch looks wrong and a specific signal is needed. An optional local, read-only MCP server lets AI coding assistants pick the diagnostic that fits a problem and insert the smallest snippet into an existing loop.

Running it means adding the diagnostic objects to a PyTorch or Hugging Face training loop. GradientHealthMonitor is called after backward() and before optimizer.step(). The example trackers need stable dataset ids, logits and labels from a normal classification run. Diagnostic results can be written to versioned JSON.

Limitations and requirements

The project is alpha software. Thresholds are heuristics and should be read as training signals, not automatic truth. The MCP server runs locally and is read-only.

Safety

Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.

  • Runs code or changes an assistant's behaviour

Sources

  • Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
    TrainTools

    Recommend paper-backed diagnostics for PyTorch and Hugging Face training problems.

    v0.6.2
    As shown on registry.modelcontextprotocol.io when we last read it
MCP servers
cloudflare

Cloudflare MCP servers

Cloudflare's hosted MCP servers for its products: documentation search, sandbox containers, browser rendering, Radar internet data, logs and analytics, Workers and more.

4.5K GitHub starsCloud & DevOpsSoftware development
Updated 17 Nov 2026
MCP servers
redis

Redis MCP Server

Redis's official MCP server that lets an AI assistant store, read and search data in a Redis database, including streams, JSON documents and vector indexes.

642 GitHub starsData & databasesSoftware development
Updated 17 Nov 2026
MCP servers
microsoft

Playwright MCP (browser automation)

Microsoft's MCP server that lets an AI assistant open web pages, click, type and read them through a real browser, using the page structure instead of screenshots.

39.6K GitHub starsTesting & QAWeb, search & browser
Updated 17 Nov 2026
MCP servers
hashicorp

Terraform MCP Server (HashiCorp)

HashiCorp's MCP server that gives an assistant current Terraform Registry information on providers, modules and policies, and can manage HCP Terraform workspaces.

1.6K GitHub starsCloud & DevOpsSoftware development
Updated 17 Nov 2026