spec: add line-loss-training specification (#2868)

## New Specification: `line-loss-training`

Related to #2860

---

### specification.md

# line-loss-training: Training Loss Curve

## Description

A line plot showing training and validation loss curves over epochs
during neural network training. This visualization is essential for
monitoring model training, detecting overfitting (when validation loss
diverges from training loss), and determining optimal early stopping
points. The dual-curve display reveals the gap between training and
generalization performance.

## Applications

- Monitoring deep learning model training to detect overfitting or
underfitting
- Determining the optimal epoch for early stopping based on validation
loss plateau or increase
- Comparing training dynamics across different model architectures or
hyperparameter configurations

## Data

- `epoch` (numeric) - Training epoch number or iteration count
- `train_loss` (numeric) - Loss values on training data at each epoch
- `val_loss` (numeric) - Loss values on validation data at each epoch
- Size: 10-500 epochs typical, logged at each epoch or at regular
intervals
- Example: Training history from Keras/PyTorch containing epoch,
training loss, and validation loss

## Notes

- Use distinct colors for training (e.g., blue) and validation (e.g.,
orange) curves
- Include a legend clearly labeling each curve
- X-axis should start at epoch 1 (or 0) and extend to final epoch
- Y-axis label should specify the loss function used (e.g.,
Cross-Entropy Loss, MSE)
- Optionally mark the epoch with minimum validation loss to indicate
optimal stopping point
- Consider log scale for y-axis if loss spans multiple orders of
magnitude

---

**Next:** Add `approved` label to the issue to merge this PR.

---
:robot: *[spec-create
workflow](https://github.com/MarkusNeusinger/pyplots/actions/runs/20608362596)*

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

Author: github-actions[bot]

plots/line-loss-training/specification.md

@@ -0,0 +1,28 @@
+# line-loss-training: Training Loss Curve
+
+## Description
+
+A line plot showing training and validation loss curves over epochs during neural network training. This visualization is essential for monitoring model training, detecting overfitting (when validation loss diverges from training loss), and determining optimal early stopping points. The dual-curve display reveals the gap between training and generalization performance.
+
+## Applications
+
+- Monitoring deep learning model training to detect overfitting or underfitting
+- Determining the optimal epoch for early stopping based on validation loss plateau or increase
+- Comparing training dynamics across different model architectures or hyperparameter configurations
+
+## Data
+
+- `epoch` (numeric) - Training epoch number or iteration count
+- `train_loss` (numeric) - Loss values on training data at each epoch
+- `val_loss` (numeric) - Loss values on validation data at each epoch
+- Size: 10-500 epochs typical, logged at each epoch or at regular intervals
+- Example: Training history from Keras/PyTorch containing epoch, training loss, and validation loss
+
+## Notes
+
+- Use distinct colors for training (e.g., blue) and validation (e.g., orange) curves
+- Include a legend clearly labeling each curve
+- X-axis should start at epoch 1 (or 0) and extend to final epoch
+- Y-axis label should specify the loss function used (e.g., Cross-Entropy Loss, MSE)
+- Optionally mark the epoch with minimum validation loss to indicate optimal stopping point
+- Consider log scale for y-axis if loss spans multiple orders of magnitude

plots/line-loss-training/specification.yaml

@@ -0,0 +1,28 @@
+# Specification-level metadata for line-loss-training
+# Auto-synced to PostgreSQL on push to main
+
+spec_id: line-loss-training
+title: Training Loss Curve
+
+# Specification tracking
+created: 2025-12-30T23:43:50Z
+updated: 2025-12-30T23:43:50Z
+issue: 2860
+suggested: MarkusNeusinger
+
+# Classification tags (applies to all library implementations)
+# See docs/concepts/tagging-system.md for detailed guidelines
+tags:
+  plot_type:
+    - line
+    - curve
+  data_type:
+    - numeric
+    - timeseries
+  domain:
+    - machine-learning
+    - data-science
+  features:
+    - multi
+    - comparison
+    - training-monitoring