Skip to main content
The monitoring dashboard gives you real-time visibility into your training runs — metrics, logs, checkpoints, and model performance over time.

Metrics Dashboard

During an active training run, the dashboard displays key metrics updated in real time:
MetricDescription
Training RewardAverage reward score across training rollouts
Validation RewardReward score on held-out validation data
Model EntropyMeasure of output diversity (higher = more exploration)
Response LengthAverage token count of model responses
KL DivergenceDistance from the reference model (monitors catastrophic forgetting)
Each metric is plotted over training steps, so you can see trends and identify when the model is converging, overfitting, or degrading.

Time Range Controls

Filter metrics by time range:
  • Last hour, last 6 hours, last 24 hours
  • Full training run
  • Custom range

Training Logs

The logs panel shows detailed event-level information:
  • Rollout logs — Individual rollout traces with prompts, responses, tool calls, and rewards
  • System logs — Infrastructure events (GPU allocation, checkpoint saves, errors)
  • Reward breakdowns — Per-sample reward scores with details from each reward function
Logs can be filtered by severity (info, warning, error) and searched by keyword.

Checkpoints

Checkpoints are saved automatically during training at configurable intervals. Each checkpoint captures:
  • Model weights at that training step
  • Training metrics at the time of save
  • Configuration used for the run

Viewing Checkpoints

The checkpoint list shows:
  • Step number — Training step when the checkpoint was saved
  • Training reward — Average reward at that step
  • Validation reward — Validation score at that step
  • Timestamp — When the checkpoint was created

Merging Checkpoints

To create a deployable model from a checkpoint:
  1. Select a checkpoint from the list
  2. Click Merge — this combines the RL adapter with the base model
  3. The merged model is saved and ready for export

Exporting to Hugging Face

After merging a checkpoint:
  1. Click Upload to Hugging Face on the merged model
  2. Configure the target repository and visibility (public/private)
  3. The model is uploaded with a model card containing training metadata
You must configure your Hugging Face integration in Workspace Settings before exporting.

Comparing Runs

If you have multiple training runs in a project, you can compare their metrics side-by-side to identify which configuration performs best.

Next Steps