TrainLens Report

Model detected: FineTunedTransformer

Framework: huggingface

Training summary

Result explanation

Metrics

MetricValue
train_accuracy0.914
train_loss0.214
validation_accuracy0.887
validation_loss0.291

Execution trace

StepEpochEventMetrics
1001.00evalvalidation_loss=0.412, validation_accuracy=0.801
5005.00evalvalidation_loss=0.291, validation_accuracy=0.887

Potential issues

Top features

1. token_overlap

2. sequence_length

3. domain_keyword_count

Recommended next steps

Improvement plan

1. Run a validation error analysis. Confidence: 82%. Why: Inspect false positives and false negatives before changing architecture.

2. Try a lower learning rate for two extra epochs. Confidence: 64%. Why: The curve suggests room for small convergence gains.