feat(matplotlib): implement line-loss-training (#2905)

## Implementation: `line-loss-training` - matplotlib

Implements the **matplotlib** version of `line-loss-training`.

**File:** `plots/line-loss-training/implementations/matplotlib.py`

**Parent Issue:** #2860

---
:robot: *[impl-generate
workflow](https://github.com/MarkusNeusinger/pyplots/actions/runs/20608625949)*

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Author: github-actions[bot]

plots/line-loss-training/implementations/matplotlib.py

@@ -0,0 +1,60 @@
+""" pyplots.ai
+line-loss-training: Training Loss Curve
+Library: matplotlib 3.10.8 | Python 3.13.11
+Quality: 94/100 | Created: 2025-12-31
+"""
+
+import matplotlib.pyplot as plt
+import numpy as np
+
+
+# Data - Simulate training and validation loss over epochs
+np.random.seed(42)
+epochs = np.arange(1, 51)
+
+# Training loss: exponential decay with small noise
+train_loss = 2.5 * np.exp(-0.08 * epochs) + 0.15 + np.random.randn(50) * 0.015
+
+# Validation loss: decays similarly but starts overfitting after epoch 28
+val_base = 2.5 * np.exp(-0.065 * epochs) + 0.30
+noise = np.random.randn(50) * 0.02
+val_loss = val_base + noise
+# Clear overfitting after epoch 28: validation loss increases while training keeps dropping
+val_loss[28:] = val_loss[28] + np.linspace(0, 0.35, 22) + np.random.randn(22) * 0.015
+
+# Find minimum validation loss epoch for annotation
+min_val_epoch = np.argmin(val_loss) + 1
+min_val_loss = val_loss[min_val_epoch - 1]
+
+# Create plot
+fig, ax = plt.subplots(figsize=(16, 9))
+
+# Plot training and validation loss curves
+ax.plot(epochs, train_loss, linewidth=3, color="#306998", label="Training Loss", marker="o", markersize=6)
+ax.plot(epochs, val_loss, linewidth=3, color="#FFD43B", label="Validation Loss", marker="s", markersize=6)
+
+# Mark minimum validation loss point (optimal early stopping)
+ax.scatter([min_val_epoch], [min_val_loss], s=300, color="#E74C3C", zorder=5, edgecolors="black", linewidth=2)
+ax.annotate(
+    f"Best: Epoch {min_val_epoch}",
+    xy=(min_val_epoch, min_val_loss),
+    xytext=(min_val_epoch - 12, min_val_loss + 0.35),
+    fontsize=16,
+    arrowprops={"arrowstyle": "->", "color": "black", "lw": 2},
+    bbox={"boxstyle": "round,pad=0.3", "facecolor": "white", "edgecolor": "black"},
+)
+
+# Labels and styling
+ax.set_xlabel("Epoch", fontsize=20)
+ax.set_ylabel("Cross-Entropy Loss", fontsize=20)
+ax.set_title("line-loss-training · matplotlib · pyplots.ai", fontsize=24)
+ax.tick_params(axis="both", labelsize=16)
+ax.legend(fontsize=16, loc="upper right")
+ax.grid(True, alpha=0.3, linestyle="--")
+
+# Set axis limits
+ax.set_xlim(0, 52)
+ax.set_ylim(0, 2.8)
+
+plt.tight_layout()
+plt.savefig("plot.png", dpi=300, bbox_inches="tight")

plots/line-loss-training/metadata/matplotlib.yaml

@@ -0,0 +1,28 @@
+library: matplotlib
+specification_id: line-loss-training
+created: '2025-12-31T00:07:24Z'
+updated: '2025-12-31T00:11:39Z'
+generated_by: claude-opus-4-5-20251101
+workflow_run: 20608625949
+issue: 2860
+python_version: 3.13.11
+library_version: 3.10.8
+preview_url: https://storage.googleapis.com/pyplots-images/plots/line-loss-training/matplotlib/plot.png
+preview_thumb: https://storage.googleapis.com/pyplots-images/plots/line-loss-training/matplotlib/plot_thumb.png
+preview_html: null
+quality_score: 94
+review:
+  strengths:
+  - Excellent demonstration of overfitting pattern with clear divergence between training
+    and validation loss
+  - Optimal early stopping point annotation is well-executed with arrow and styled
+    bbox
+  - Color choices (blue and yellow) are colorblind-safe and provide excellent contrast
+  - Clean KISS structure with proper random seed for reproducibility
+  - All specification requirements are met including distinct colors, legend, and
+    loss function labeling
+  weaknesses:
+  - Legend position in upper right slightly overlaps with early epoch data points;
+    consider upper center or outside plot
+  - Could use matplotlib-specific features like fill_between for confidence intervals
+    or secondary annotations