- LIVE DATA
- ANALYSIS
- REPORTING
- HUMAN
Short answer
How can I check my training metrics using AI?
Your AI uses the Neptune.ai connector to pull live data from your active experiments. You can ask for specific accuracy scores, loss curves, or hyperparameter settings from any run. This gives you a direct line into your experiment logs through natural language.
The outcomes
What you get from your metrics
The AI turns raw experiment data into actionable insights.
LIVE DATA
Track active runs
The AI pulls current metrics from ongoing experiments. You get real-time updates on how your model is performing right now.
ANALYSIS
Compare hyperparameters
Ask your agent to look at specific settings used in different runs. It identifies which configurations led to better results.
REPORTING
Summarize performance
Get a quick rundown of accuracy and loss trends. The AI condenses complex logs into a clear summary of model health.
HUMAN
Decide next steps
The AI provides the data, but you decide if a run needs to be stopped or if you should adjust your learning rate.
The workflow
What your AI does when the data arrives.
The AI acts as a bridge between your Neptune.ai dashboard and your conversation.
Find the right run
The AI identifies the specific experiment or run you are interested in by looking through your project history.
search_runsExtract specific values
Once the run is found, the AI pulls the exact metric or parameter value you asked for.
get_attributesCompare results
The AI looks at multiple runs to find patterns in how different settings affect your loss or accuracy.
search_runsget_attributesPresent findings
The AI translates the raw numbers into a plain text summary that answers your specific question.
get_attributes
Try it
Copy these to start.
Use these prompts to begin interacting with your experiment data.
Starting points
Replace the bracketed text with your actual Neptune.ai project names or run IDs.
Neptune.ai Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_HI0D1zHVFTWAvCQH47Rd6lplgg0GBJXVQZ0fMTHm/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Neptune.ai capabilities are ready to use.
{
"mcpServers": {
"neptuneai-ml-experiment-tracking-mcp": {
"url": "https://edge.vinkius.com/vk_preview_HI0D1zHVFTWAvCQH47Rd6lplgg0GBJXVQZ0fMTHm/mcp"
}
}
}Show me the current accuracy for the run named [Run ID].
What were the hyperparameters used in the [Project Name] experiment?
Compare the loss values between [Run A] and [Run B].
Find the run in [Project Name] that had the highest validation accuracy.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Neptune.ai once, then ask.
Follow the link to authorize the connection. Your credentials stay encrypted and secure, and you can immediately start asking questions about your runs.
Connect Neptune.ai to your AIFAQ
How this task behaves
- 01
Can the AI change my hyperparameter settings?
No. The AI can only read data from Neptune.ai. You must manually update your code or configuration to change settings.
- 02
Can the AI delete old experiment runs?
No. This connector is read-only. It cannot delete, rename, or modify any data in your Neptune.ai account.
- 03
Does the AI see all my projects?
The AI can access any run it can find through the search tools, provided you have given it access to the relevant project.
- 04
Can I get plots and charts?
The AI retrieves the raw numerical data for metrics. It can describe the trends it sees, but it does not render visual graphs.
- 05
How does it find specific runs?
It uses the search tools to scan your project for runs that match the names or attributes you mention.
More questions about Neptune.ai? See everything the Neptune.ai Connector can do
Connect Neptune.ai to Claude, Cursor, ChatGPT & more
