Add loss visualization

lnp_ml/modeling/pretrain.py

@@ -14,6 +14,7 @@ import typer
 
 from lnp_ml.config import MODELS_DIR, PROCESSED_DATA_DIR
 from lnp_ml.dataset import ExternalDeliveryDataset, collate_fn
+from lnp_ml.modeling.visualization import plot_loss_curves
 
 # MPNN ensemble 默认路径
 DEFAULT_MPNN_ENSEMBLE_DIR = MODELS_DIR / "mpnn" / "all_amine_split_for_LiON"
@@ -351,6 +352,15 @@ def main(
         json.dump(result["history"], f, indent=2)
     logger.success(f"Saved pretrain history to {history_path}")
 
+    # 绘制 loss 曲线图
+    loss_plot_path = output_dir / "pretrain_loss_curves.png"
+    plot_loss_curves(
+        history=result["history"],
+        output_path=loss_plot_path,
+        title="Pretrain Loss Curves (Delivery)",
+    )
+    logger.success(f"Saved loss curves plot to {loss_plot_path}")
+
     logger.success(
         f"Pretraining complete! Best val_loss: {result['best_val_loss']:.4f}"
     )

lnp_ml/modeling/pretrain_cv.py

@@ -16,6 +16,7 @@ import typer
 
 from lnp_ml.config import MODELS_DIR, PROCESSED_DATA_DIR
 from lnp_ml.dataset import ExternalDeliveryDataset, collate_fn
+from lnp_ml.modeling.visualization import plot_loss_curves
 
 
 # MPNN ensemble 默认路径
@@ -226,6 +227,15 @@ def train_fold(
     with open(history_path, "w") as f:
         json.dump(history, f, indent=2)
     
+    # 绘制 loss 曲线图
+    loss_plot_path = fold_output_dir / "loss_curves.png"
+    plot_loss_curves(
+        history=history,
+        output_path=loss_plot_path,
+        title=f"Pretrain Fold {fold_idx} Loss Curves",
+    )
+    logger.info(f"Saved fold {fold_idx} loss curves to {loss_plot_path}")
+    
     return {
         "fold_idx": fold_idx,
         "best_val_loss": best_val_loss,

lnp_ml/modeling/train.py

@@ -19,6 +19,7 @@ from lnp_ml.modeling.trainer import (
     EarlyStopping,
     LossWeights,
 )
+from lnp_ml.modeling.visualization import plot_multitask_loss_curves
 
 # MPNN ensemble 默认路径
 DEFAULT_MPNN_ENSEMBLE_DIR = MODELS_DIR / "mpnn" / "all_amine_split_for_LiON"
@@ -406,6 +407,15 @@ def main(
         json.dump(result["history"], f, indent=2)
     logger.success(f"Saved training history to {history_path}")
     
+    # 绘制多任务 loss 曲线图
+    loss_plot_path = output_dir / "loss_curves.png"
+    plot_multitask_loss_curves(
+        history=result["history"],
+        output_path=loss_plot_path,
+        title="Multi-task Training Loss Curves",
+    )
+    logger.success(f"Saved loss curves plot to {loss_plot_path}")
+    
     logger.success(f"Training complete! Best val_loss: {result['best_val_loss']:.4f}")
 
 

lnp_ml/modeling/train_cv.py

@@ -22,6 +22,7 @@ from lnp_ml.modeling.trainer import (
     EarlyStopping,
     LossWeights,
 )
+from lnp_ml.modeling.visualization import plot_multitask_loss_curves
 
 
 # MPNN ensemble 默认路径
@@ -158,6 +159,15 @@ def train_fold(
     with open(history_path, "w") as f:
         json.dump(history, f, indent=2)
     
+    # 绘制多任务 loss 曲线图
+    loss_plot_path = fold_output_dir / "loss_curves.png"
+    plot_multitask_loss_curves(
+        history=history,
+        output_path=loss_plot_path,
+        title=f"Fold {fold_idx} Multi-task Loss Curves",
+    )
+    logger.info(f"Saved fold {fold_idx} loss curves to {loss_plot_path}")
+    
     return {
         "fold_idx": fold_idx,
         "best_val_loss": best_val_loss,

lnp_ml/modeling/trainer.py

@@ -13,12 +13,18 @@ from tqdm import tqdm
 @dataclass
 class LossWeights:
     """各任务的损失权重"""
-    size: float = 1.0
+    # size: float = 1.0
-    pdi: float = 1.0
+    # pdi: float = 1.0
-    ee: float = 1.0
+    # ee: float = 1.0
+    # delivery: float = 1.0
+    # biodist: float = 1.0
+    # toxic: float = 1.0
+    size: float = 0.1
+    pdi: float = 0.3
+    ee: float = 0.3
     delivery: float = 1.0
     biodist: float = 1.0
-    toxic: float = 1.0
+    toxic: float = 0.05
 
 
 def compute_multitask_loss(

lnp_ml/modeling/visualization.py

@@ -0,0 +1,284 @@
+"""训练过程可视化工具：绘制 loss 曲线"""
+
+from pathlib import Path
+from typing import Dict, List, Optional, Union
+import matplotlib.pyplot as plt
+import matplotlib
+matplotlib.use('Agg')  # 使用非交互式后端，避免 GUI 依赖
+
+# 设置中文字体支持
+plt.rcParams['font.sans-serif'] = ['Arial Unicode MS', 'SimHei', 'DejaVu Sans']
+plt.rcParams['axes.unicode_minus'] = False
+
+
+def plot_loss_curves(
+    history: Union[Dict[str, List[Dict]], List[Dict]],
+    output_path: Path,
+    title: str = "Training Loss Curves",
+    figsize: tuple = (12, 8),
+) -> None:
+    """
+    绘制训练过程中各个 loss 组成部分的变化曲线。
+
+    支持两种 history 格式：
+    1. 预训练格式（单任务）：
+       {"train": [{"loss": 0.1, ...}, ...], "val": [{"loss": 0.1, ...}, ...]}
+    2. CV fold 格式：
+       [{"epoch": 1, "train_loss": 0.1, "val_loss": 0.1, ...}, ...]
+
+    Args:
+        history: 训练历史记录
+        output_path: 输出图片路径
+        title: 图标题
+        figsize: 图片尺寸
+    """
+    output_path = Path(output_path)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    # 判断 history 格式并统一处理
+    if isinstance(history, dict) and "train" in history:
+        # 格式1: {"train": [...], "val": [...]}
+        _plot_standard_history(history, output_path, title, figsize)
+    elif isinstance(history, list):
+        # 格式2: [{...}, {...}, ...]
+        _plot_flat_history(history, output_path, title, figsize)
+    else:
+        raise ValueError(f"Unsupported history format: {type(history)}")
+
+
+def _plot_standard_history(
+    history: Dict[str, List[Dict]],
+    output_path: Path,
+    title: str,
+    figsize: tuple,
+) -> None:
+    """绘制标准格式的 history（train/val 分开）"""
+    train_history = history.get("train", [])
+    val_history = history.get("val", [])
+
+    if not train_history:
+        return
+
+    epochs = list(range(1, len(train_history) + 1))
+
+    # 收集所有 loss 键
+    all_loss_keys = set()
+    for record in train_history + val_history:
+        for key in record.keys():
+            if key.startswith("loss") or key == "loss":
+                all_loss_keys.add(key)
+
+    # 分离总 loss 和各任务 loss
+    total_loss_key = "loss"
+    task_loss_keys = sorted([k for k in all_loss_keys if k != "loss"])
+
+    # 创建子图
+    n_subplots = 1 + (1 if task_loss_keys else 0)
+    fig, axes = plt.subplots(n_subplots, 1, figsize=(figsize[0], figsize[1] * n_subplots / 2))
+    if n_subplots == 1:
+        axes = [axes]
+
+    # 颜色配置
+    colors = plt.cm.tab10.colors
+
+    # 子图1：总 loss
+    ax = axes[0]
+    train_total_loss = [r.get(total_loss_key, 0) for r in train_history]
+    val_total_loss = [r.get(total_loss_key, 0) for r in val_history]
+
+    ax.plot(epochs, train_total_loss, 'o-', label='Train Total Loss', color=colors[0], linewidth=2, markersize=4)
+    if val_total_loss:
+        ax.plot(epochs, val_total_loss, 's--', label='Val Total Loss', color=colors[1], linewidth=2, markersize=4)
+
+    ax.set_xlabel('Epoch', fontsize=12)
+    ax.set_ylabel('Loss', fontsize=12)
+    ax.set_title(f'{title} - Total Loss', fontsize=14, fontweight='bold')
+    ax.legend(loc='upper right', fontsize=10)
+    ax.grid(True, alpha=0.3)
+    ax.set_xlim(0.5, len(epochs) + 0.5)
+
+    # 子图2：各任务 loss（如果有）
+    if task_loss_keys:
+        ax = axes[1]
+        for i, key in enumerate(task_loss_keys):
+            task_name = key.replace("loss_", "").upper()
+            train_values = [r.get(key, 0) for r in train_history]
+            val_values = [r.get(key, 0) for r in val_history]
+
+            color = colors[i % len(colors)]
+            ax.plot(epochs, train_values, 'o-', label=f'Train {task_name}', color=color, alpha=0.8, linewidth=1.5, markersize=3)
+            if val_values and any(v > 0 for v in val_values):
+                ax.plot(epochs, val_values, 's--', label=f'Val {task_name}', color=color, alpha=0.5, linewidth=1.5, markersize=3)
+
+        ax.set_xlabel('Epoch', fontsize=12)
+        ax.set_ylabel('Loss', fontsize=12)
+        ax.set_title(f'{title} - Per-Task Loss', fontsize=14, fontweight='bold')
+        ax.legend(loc='upper right', fontsize=9, ncol=2)
+        ax.grid(True, alpha=0.3)
+        ax.set_xlim(0.5, len(epochs) + 0.5)
+
+    plt.tight_layout()
+    plt.savefig(output_path, dpi=150, bbox_inches='tight', facecolor='white')
+    plt.close(fig)
+
+
+def _plot_flat_history(
+    history: List[Dict],
+    output_path: Path,
+    title: str,
+    figsize: tuple,
+) -> None:
+    """绘制扁平格式的 history（CV fold 格式）"""
+    if not history:
+        return
+
+    epochs = [r.get("epoch", i + 1) for i, r in enumerate(history)]
+
+    # 收集所有 loss 相关的键
+    loss_keys = set()
+    for record in history:
+        for key in record.keys():
+            if "loss" in key.lower():
+                loss_keys.add(key)
+
+    # 分类
+    train_keys = sorted([k for k in loss_keys if "train" in k.lower()])
+    val_keys = sorted([k for k in loss_keys if "val" in k.lower()])
+
+    # 创建子图
+    fig, ax = plt.subplots(1, 1, figsize=figsize)
+
+    colors = plt.cm.tab10.colors
+    color_idx = 0
+
+    # 绘制训练 loss
+    for key in train_keys:
+        values = [r.get(key, 0) for r in history]
+        label = key.replace("_", " ").title()
+        ax.plot(epochs, values, 'o-', label=label, color=colors[color_idx % len(colors)], 
+                linewidth=2, markersize=4, alpha=0.9)
+        color_idx += 1
+
+    # 绘制验证 loss
+    for key in val_keys:
+        values = [r.get(key, 0) for r in history]
+        label = key.replace("_", " ").title()
+        ax.plot(epochs, values, 's--', label=label, color=colors[color_idx % len(colors)], 
+                linewidth=2, markersize=4, alpha=0.7)
+        color_idx += 1
+
+    ax.set_xlabel('Epoch', fontsize=12)
+    ax.set_ylabel('Loss', fontsize=12)
+    ax.set_title(title, fontsize=14, fontweight='bold')
+    ax.legend(loc='upper right', fontsize=10)
+    ax.grid(True, alpha=0.3)
+    ax.set_xlim(0.5, len(epochs) + 0.5)
+
+    plt.tight_layout()
+    plt.savefig(output_path, dpi=150, bbox_inches='tight', facecolor='white')
+    plt.close(fig)
+
+
+def plot_multitask_loss_curves(
+    history: Dict[str, List[Dict]],
+    output_path: Path,
+    title: str = "Multi-task Training Loss",
+    figsize: tuple = (14, 10),
+) -> None:
+    """
+    专门用于多任务训练的 loss 曲线绘制。
+    
+    将各个任务的 loss 分别绘制在不同的子图中，便于比较。
+    
+    Args:
+        history: {"train": [...], "val": [...]} 格式的训练历史
+        output_path: 输出路径
+        title: 图标题
+        figsize: 图片尺寸
+    """
+    output_path = Path(output_path)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    train_history = history.get("train", [])
+    val_history = history.get("val", [])
+
+    if not train_history:
+        return
+
+    epochs = list(range(1, len(train_history) + 1))
+
+    # 提取所有任务的 loss 键
+    task_keys = set()
+    for record in train_history:
+        for key in record.keys():
+            if key.startswith("loss_"):
+                task_name = key.replace("loss_", "")
+                task_keys.add(task_name)
+
+    task_keys = sorted(task_keys)
+
+    # 计算子图布局
+    n_tasks = len(task_keys)
+    if n_tasks == 0:
+        # 只有总 loss，使用简单绘图
+        _plot_standard_history(history, output_path, title, figsize)
+        return
+
+    # 包含总 loss，共 n_tasks + 1 个子图
+    n_plots = n_tasks + 1
+    n_cols = min(3, n_plots)
+    n_rows = (n_plots + n_cols - 1) // n_cols
+
+    fig, axes = plt.subplots(n_rows, n_cols, figsize=(figsize[0], figsize[1] * n_rows / 2))
+    if n_plots == 1:
+        axes = [[axes]]
+    elif n_rows == 1:
+        axes = [axes]
+
+    axes_flat = [ax for row in axes for ax in (row if hasattr(row, '__iter__') else [row])]
+
+    colors = plt.cm.tab10.colors
+
+    # 子图1：总 loss
+    ax = axes_flat[0]
+    train_total = [r.get("loss", 0) for r in train_history]
+    val_total = [r.get("loss", 0) for r in val_history]
+
+    ax.plot(epochs, train_total, 'o-', label='Train', color=colors[0], linewidth=2, markersize=4)
+    if val_total:
+        ax.plot(epochs, val_total, 's--', label='Val', color=colors[1], linewidth=2, markersize=4)
+    ax.set_xlabel('Epoch')
+    ax.set_ylabel('Loss')
+    ax.set_title('Total Loss', fontweight='bold')
+    ax.legend(loc='upper right')
+    ax.grid(True, alpha=0.3)
+
+    # 各任务子图
+    for idx, task in enumerate(task_keys):
+        ax = axes_flat[idx + 1]
+        key = f"loss_{task}"
+
+        train_values = [r.get(key, 0) for r in train_history]
+        val_values = [r.get(key, 0) for r in val_history]
+
+        # 只绘制有值的数据
+        if any(v > 0 for v in train_values):
+            ax.plot(epochs, train_values, 'o-', label='Train', color=colors[0], linewidth=2, markersize=4)
+        if val_values and any(v > 0 for v in val_values):
+            ax.plot(epochs, val_values, 's--', label='Val', color=colors[1], linewidth=2, markersize=4)
+
+        ax.set_xlabel('Epoch')
+        ax.set_ylabel('Loss')
+        ax.set_title(f'{task.upper()} Loss', fontweight='bold')
+        ax.legend(loc='upper right')
+        ax.grid(True, alpha=0.3)
+
+    # 隐藏多余的子图
+    for idx in range(n_plots, len(axes_flat)):
+        axes_flat[idx].set_visible(False)
+
+    plt.suptitle(title, fontsize=16, fontweight='bold', y=1.02)
+    plt.tight_layout()
+    plt.savefig(output_path, dpi=150, bbox_inches='tight', facecolor='white')
+    plt.close(fig)
+

models/finetune_cv/config.json

@@ -11,6 +11,6 @@
   "batch_size": 32,
   "epochs": 100,
   "patience": 15,
-  "init_from_pretrain": null,
+  "init_from_pretrain": "models/pretrain_delivery.pt",
   "freeze_backbone": false
 }

models/finetune_cv/cv_results.json

@@ -2,38 +2,38 @@
   "fold_results": [
     {
       "fold_idx": 0,
-      "best_val_loss": 5.7676777839660645,
+      "best_val_loss": 0.9860520362854004,
-      "epochs_trained": 24,
+      "epochs_trained": 40,
-      "final_train_loss": 1.4942118644714355
+      "final_train_loss": 0.5008097920152876
     },
     {
       "fold_idx": 1,
-      "best_val_loss": 8.418675899505615,
+      "best_val_loss": 2.4599782625834146,
-      "epochs_trained": 20,
+      "epochs_trained": 38,
-      "final_train_loss": 1.4902493238449097
+      "final_train_loss": 0.564177993271086
     },
     {
       "fold_idx": 2,
-      "best_val_loss": 3.5122547830854143,
+      "best_val_loss": 0.7660132050514221,
-      "epochs_trained": 25,
+      "epochs_trained": 43,
-      "final_train_loss": 1.7609570423762004
+      "final_train_loss": 0.6722757054699792
     },
     {
       "fold_idx": 3,
-      "best_val_loss": 3.165306806564331,
+      "best_val_loss": 1.065057098865509,
-      "epochs_trained": 21,
+      "epochs_trained": 31,
-      "final_train_loss": 2.0073827385902403
+      "final_train_loss": 0.7323974437183804
     },
     {
       "fold_idx": 4,
-      "best_val_loss": 2.996154228846232,
+      "best_val_loss": 1.321769932905833,
-      "epochs_trained": 18,
+      "epochs_trained": 36,
-      "final_train_loss": 1.9732873006300493
+      "final_train_loss": 0.5991987817817264
     }
   ],
   "summary": {
-    "val_loss_mean": 4.772013900393532,
+    "val_loss_mean": 1.3197741071383158,
-    "val_loss_std": 2.0790222989111475
+    "val_loss_std": 0.5971552245392587
   },
   "config": {
     "d_model": 256,
@@ -48,7 +48,7 @@
     "batch_size": 32,
     "epochs": 100,
     "patience": 15,
-    "init_from_pretrain": null,
+    "init_from_pretrain": "models/pretrain_delivery.pt",
     "freeze_backbone": false
   }
 }

models/finetune_cv/test_results.json

@@ -2,293 +2,293 @@
   "fold_results": [
     {
       "fold_idx": 0,
-      "n_samples": 95,
+      "n_samples": 87,
       "size": {
-        "n": 95,
+        "n": 87,
-        "rmse": 0.5909209144067168,
+        "rmse": 0.5578980261231201,
-        "mae": 0.376253614927593,
+        "mae": 0.293820194814397,
-        "r2": 0.005712927161997228
+        "r2": -0.16350852977979646
       },
       "delivery": {
-        "n": 66,
+        "n": 64,
-        "rmse": 1.3280577883458438,
+        "rmse": 1.4180242556680394,
-        "mae": 0.5195405159964028,
+        "mae": 0.489678907149937,
-        "r2": 0.03195999739694366
+        "r2": 0.006066588761596381
       },
       "pdi": {
-        "n": 95,
+        "n": 87,
-        "accuracy": 0.6105263157894737,
+        "accuracy": 0.7011494252873564,
-        "precision": 0.20350877192982456,
+        "precision": 0.382983682983683,
-        "recall": 0.3333333333333333,
+        "recall": 0.3981244671781756,
-        "f1": 0.25272331154684097
+        "f1": 0.38831483607603007
       },
       "ee": {
-        "n": 95,
+        "n": 87,
-        "accuracy": 0.6736842105263158,
+        "accuracy": 0.6206896551724138,
-        "precision": 0.22456140350877193,
+        "precision": 0.4700393567388641,
-        "recall": 0.3333333333333333,
+        "recall": 0.5084656084656085,
-        "f1": 0.26834381551362685
+        "f1": 0.4656627032698024
       },
       "toxic": {
-        "n": 66,
+        "n": 65,
-        "accuracy": 0.8939393939393939,
+        "accuracy": 0.9846153846153847,
-        "precision": 0.44696969696969696,
+        "precision": 0.9918032786885246,
-        "recall": 0.5,
+        "recall": 0.9,
-        "f1": 0.472
+        "f1": 0.9403122130394859
       },
       "biodist": {
-        "n": 66,
+        "n": 65,
-        "kl_divergence": 0.851655784204727,
+        "kl_divergence": 0.20084619060054956,
-        "js_divergence": 0.21404831573756974
+        "js_divergence": 0.048292698388857934
       }
     },
     {
       "fold_idx": 1,
-      "n_samples": 195,
+      "n_samples": 87,
       "size": {
-        "n": 193,
+        "n": 87,
-        "rmse": 0.4425801645813746,
+        "rmse": 0.40406802223294774,
-        "mae": 0.26432527161632796,
+        "mae": 0.2981993521767101,
-        "r2": -0.026225211870033682
+        "r2": 0.09717283856858072
       },
       "delivery": {
-        "n": 123,
+        "n": 58,
-        "rmse": 0.7771322048436382,
+        "rmse": 0.545857341773909,
-        "mae": 0.6133777339870822,
+        "mae": 0.3866841504832023,
-        "r2": -0.128644776760948
+        "r2": 0.39045172405838346
       },
       "pdi": {
-        "n": 195,
+        "n": 87,
-        "accuracy": 0.7076923076923077,
+        "accuracy": 0.8045977011494253,
-        "precision": 0.35384615384615387,
+        "precision": 0.49282452707110247,
-        "recall": 0.5,
+        "recall": 0.45567765567765567,
-        "f1": 0.4144144144144144
+        "f1": 0.4661145617667357
       },
       "ee": {
-        "n": 195,
+        "n": 87,
-        "accuracy": 0.4205128205128205,
+        "accuracy": 0.7241379310344828,
-        "precision": 0.14017094017094017,
+        "precision": 0.7064586357039188,
-        "recall": 0.3333333333333333,
+        "recall": 0.6503496503496503,
-        "f1": 0.19735258724428398
+        "f1": 0.6654761904761903
       },
       "toxic": {
-        "n": 123,
+        "n": 59,
-        "accuracy": 1.0,
+        "accuracy": 0.9830508474576272,
-        "precision": 1.0,
+        "precision": 0.9912280701754386,
-        "recall": 1.0,
+        "recall": 0.8333333333333333,
-        "f1": 1.0
+        "f1": 0.8955752212389381
       },
       "biodist": {
-        "n": 123,
+        "n": 58,
-        "kl_divergence": 0.9336461102028436,
+        "kl_divergence": 0.185822128176519,
-        "js_divergence": 0.24870266224462317
+        "js_divergence": 0.049566546350752166
       }
     },
     {
       "fold_idx": 2,
-      "n_samples": 51,
+      "n_samples": 87,
       "size": {
-        "n": 51,
+        "n": 86,
-        "rmse": 0.6473513298834871,
+        "rmse": 0.5861093745094258,
-        "mae": 0.5600235602434944,
+        "mae": 0.35274335949919944,
-        "r2": -9.27515642706235
+        "r2": -0.3079648452189143
       },
       "delivery": {
-        "n": 44,
+        "n": 61,
-        "rmse": 0.7721077356414991,
+        "rmse": 0.5034529339588798,
-        "mae": 0.6167582499593581,
+        "mae": 0.3725305872618175,
-        "r2": -0.4822886602727561
+        "r2": 0.596618413667312
       },
       "pdi": {
-        "n": 51,
+        "n": 87,
-        "accuracy": 0.8823529411764706,
+        "accuracy": 0.7701149425287356,
-        "precision": 0.29411764705882354,
+        "precision": 0.7266666666666666,
-        "recall": 0.3333333333333333,
+        "recall": 0.6349206349206349,
-        "f1": 0.3125
+        "f1": 0.6497584541062802
       },
       "ee": {
-        "n": 51,
+        "n": 87,
-        "accuracy": 0.8431372549019608,
+        "accuracy": 0.5172413793103449,
-        "precision": 0.28104575163398693,
+        "precision": 0.43155828639699606,
-        "recall": 0.3333333333333333,
+        "recall": 0.40000766812361016,
-        "f1": 0.3049645390070922
+        "f1": 0.3980599647266314
       },
       "toxic": {
-        "n": 47,
+        "n": 61,
-        "accuracy": 0.851063829787234,
+        "accuracy": 1.0,
-        "precision": 0.425531914893617,
+        "precision": 1.0,
-        "recall": 0.5,
+        "recall": 1.0,
-        "f1": 0.4597701149425288
+        "f1": 1.0
       },
       "biodist": {
-        "n": 45,
+        "n": 61,
-        "kl_divergence": 1.1049896129018548,
+        "kl_divergence": 0.2646404257700098,
-        "js_divergence": 0.25485248115851133
+        "js_divergence": 0.07024299955112
       }
     },
     {
       "fold_idx": 3,
-      "n_samples": 66,
+      "n_samples": 86,
       "size": {
-        "n": 66,
+        "n": 86,
-        "rmse": 0.2407212117920812,
+        "rmse": 0.32742961478246685,
-        "mae": 0.19363613562150436,
+        "mae": 0.25193805472795355,
-        "r2": -0.11204941379936861
+        "r2": -0.09589933555096875
       },
       "delivery": {
-        "n": 62,
+        "n": 68,
-        "rmse": 1.0041711455927012,
+        "rmse": 0.7277366648519259,
-        "mae": 0.7132550483914993,
+        "mae": 0.42998586144462664,
-        "r2": -0.63265374674746
+        "r2": 0.4053674615039361
       },
       "pdi": {
-        "n": 66,
+        "n": 86,
-        "accuracy": 0.8484848484848485,
+        "accuracy": 0.7906976744186046,
-        "precision": 0.42424242424242425,
+        "precision": 0.7513157894736842,
-        "recall": 0.5,
+        "recall": 0.6356534090909091,
-        "f1": 0.4590163934426229
+        "f1": 0.6544642857142857
       },
       "ee": {
-        "n": 66,
+        "n": 86,
-        "accuracy": 0.8181818181818182,
+        "accuracy": 0.7441860465116279,
-        "precision": 0.27692307692307694,
+        "precision": 0.6962905144216474,
-        "recall": 0.32727272727272727,
+        "recall": 0.6327243018419488,
-        "f1": 0.3
+        "f1": 0.6557768628760061
       },
       "toxic": {
-        "n": 62,
+        "n": 68,
         "accuracy": 1.0,
         "precision": 1.0,
         "recall": 1.0,
         "f1": 1.0
       },
       "biodist": {
-        "n": 62,
+        "n": 68,
-        "kl_divergence": 0.9677978984139058,
+        "kl_divergence": 0.3411994760293469,
-        "js_divergence": 0.2020309307244639
+        "js_divergence": 0.07812197338009717
       }
     },
     {
       "fold_idx": 4,
-      "n_samples": 27,
+      "n_samples": 87,
       "size": {
-        "n": 27,
+        "n": 86,
-        "rmse": 0.23392834445509142,
+        "rmse": 0.28795189907825647,
-        "mae": 0.19066280788845485,
+        "mae": 0.21156823080639506,
-        "r2": -0.2667651950955112
+        "r2": 0.2077479731717109
       },
       "delivery": {
-        "n": 15,
+        "n": 59,
-        "rmse": 1.9603892288630869,
+        "rmse": 0.8048179107025805,
-        "mae": 1.3892907698949177,
+        "mae": 0.5188011898327682,
-        "r2": -0.29760739742916287
+        "r2": 0.24048521206149798
       },
       "pdi": {
-        "n": 27,
+        "n": 87,
-        "accuracy": 0.8888888888888888,
+        "accuracy": 0.6896551724137931,
-        "precision": 0.4444444444444444,
+        "precision": 0.4101075268817204,
-        "recall": 0.5,
+        "recall": 0.425,
-        "f1": 0.47058823529411764
+        "f1": 0.4174194267871083
       },
       "ee": {
-        "n": 27,
+        "n": 87,
-        "accuracy": 0.5925925925925926,
+        "accuracy": 0.7011494252873564,
-        "precision": 0.19753086419753085,
+        "precision": 0.6600529100529101,
-        "recall": 0.3333333333333333,
+        "recall": 0.581219806763285,
-        "f1": 0.24806201550387597
+        "f1": 0.5953238953238954
       },
       "toxic": {
-        "n": 15,
+        "n": 60,
-        "accuracy": 1.0,
+        "accuracy": 0.95,
-        "precision": 1.0,
+        "precision": 0.8240740740740741,
-        "recall": 1.0,
+        "recall": 0.8818181818181818,
-        "f1": 1.0
+        "f1": 0.8498748957464553
       },
       "biodist": {
-        "n": 15,
+        "n": 59,
-        "kl_divergence": 0.9389607012315264,
+        "kl_divergence": 0.207699088365002,
-        "js_divergence": 0.2470218476598176
+        "js_divergence": 0.05288953180347253
       }
     }
   ],
   "summary_stats": {
     "size": {
-      "rmse_mean": 0.43110039302375025,
+      "rmse_mean": 0.43269138734524343,
-      "rmse_std": 0.17179051271013462,
+      "rmse_std": 0.12005218734930377,
-      "r2_mean": -1.9348966641330534,
+      "r2_mean": -0.05249037976187758,
-      "r2_std": 3.6713441784129
+      "r2_std": 0.18417364026118202
     },
     "delivery": {
-      "rmse_mean": 1.1683716206573538,
+      "rmse_mean": 0.7999778213910669,
-      "rmse_std": 0.4449374578352648,
+      "rmse_std": 0.3285507063187239,
-      "r2_mean": -0.30184691676267666,
+      "r2_mean": 0.32779788001054516,
-      "r2_std": 0.23809090378746706
+      "r2_std": 0.19664259417310184
     },
     "pdi": {
-      "accuracy_mean": 0.7875890604063979,
+      "accuracy_mean": 0.751242983159583,
-      "accuracy_std": 0.11016791908756088,
+      "accuracy_std": 0.04703609766404652,
-      "f1_mean": 0.3818484709395992,
+      "f1_mean": 0.515214312890088,
-      "f1_std": 0.08529090446864619
+      "f1_std": 0.11451705845950987
     },
     "ee": {
-      "accuracy_mean": 0.6696217393431015,
+      "accuracy_mean": 0.6614808874632452,
-      "accuracy_std": 0.15503740047242787,
+      "accuracy_std": 0.08343692429542732,
-      "f1_mean": 0.2637445914537758,
+      "f1_mean": 0.5560599233345052,
-      "f1_std": 0.039213602228007696
+      "f1_std": 0.10638861908930271
     },
     "toxic": {
-      "accuracy_mean": 0.9490006447453256,
+      "accuracy_mean": 0.9835332464146024,
-      "accuracy_std": 0.06391582554207781,
+      "accuracy_std": 0.018265761928954075,
-      "f1_mean": 0.7863540229885058,
+      "f1_mean": 0.9371524660049758,
-      "f1_std": 0.26169039387919035
+      "f1_std": 0.05874632007606866
     },
     "biodist": {
-      "kl_mean": 0.9594100213909715,
+      "kl_mean": 0.24004146178828548,
-      "kl_std": 0.08240959093662605,
+      "kl_std": 0.05720155140134621,
-      "js_mean": 0.23333124750499712,
+      "js_mean": 0.05982274989485996,
-      "js_std": 0.021158533549255752
+      "js_std": 0.012080103435264247
     }
   },
   "overall": {
     "size": {
       "n_samples": 432,
-      "mse": 0.22604480336185886,
+      "mse": 0.20179930058631468,
-      "rmse": 0.47544169291497657,
+      "rmse": 0.44922077043065883,
-      "mae": 0.3084443360567093,
+      "mae": 0.2817203010673876,
-      "r2": -0.2313078534105617
+      "r2": -0.09923811531698323
     },
     "delivery": {
       "n_samples": 310,
-      "mse": 1.0873755440675295,
+      "mse": 0.760202623547859,
-      "rmse": 1.0427730069710903,
+      "rmse": 0.8718959935381393,
-      "mae": 0.6513989447841361,
+      "mae": 0.4398058238289049,
-      "r2": -0.09443640807387799
+      "r2": 0.23486100707044422
     },
     "pdi": {
       "n_samples": 434,
-      "accuracy": 0.7396313364055299,
+      "accuracy": 0.7511520737327189,
-      "precision": 0.18490783410138248,
+      "precision": 0.3287852263755878,
-      "recall": 0.25,
+      "recall": 0.3184060228452752,
-      "f1": 0.21258278145695364
+      "f1": 0.3212571677885814
     },
     "ee": {
       "n_samples": 434,
-      "accuracy": 0.5967741935483871,
+      "accuracy": 0.6612903225806451,
-      "precision": 0.1993841416474211,
+      "precision": 0.5836247086247086,
-      "recall": 0.33205128205128204,
+      "recall": 0.5469842657342657,
-      "f1": 0.24915824915824913
+      "f1": 0.5597637622559741
     },
     "toxic": {
       "n_samples": 313,
-      "accuracy": 0.9552715654952076,
+      "accuracy": 0.9840255591054313,
-      "precision": 0.4776357827476038,
+      "precision": 0.918076923076923,
-      "recall": 0.5,
+      "recall": 0.8895126612517916,
-      "f1": 0.48856209150326796
+      "f1": 0.9032337847028998
     },
     "biodist": {
       "n_samples": 311,
-      "kl_divergence": 0.9481034280166569,
+      "kl_divergence": 0.24254521665201,
-      "js_divergence": 0.23285280825310384
+      "js_divergence": 0.06022985409160514
     }
   }
 }

models/pretrain_history.json

@@ -1,205 +1,293 @@
 {
   "train": [
     {
-      "loss": 0.7730368412685099,
+      "loss": 0.7867247516051271,
       "n_samples": 6783
     },
     {
-      "loss": 0.658895703010919,
+      "loss": 0.6515523084370274,
       "n_samples": 6783
     },
     {
-      "loss": 0.6059015260392299,
+      "loss": 0.5990842743185651,
       "n_samples": 6783
     },
     {
-      "loss": 0.5744731174349416,
+      "loss": 0.5633418128920326,
       "n_samples": 6783
     },
     {
-      "loss": 0.5452056020458733,
+      "loss": 0.5453761521296815,
       "n_samples": 6783
     },
     {
-      "loss": 0.5138543470936083,
+      "loss": 0.49953126002250825,
       "n_samples": 6783
     },
     {
-      "loss": 0.4885380559178135,
+      "loss": 0.49147369265204843,
       "n_samples": 6783
     },
     {
-      "loss": 0.47587182296687974,
+      "loss": 0.4659397622863399,
       "n_samples": 6783
     },
     {
-      "loss": 0.4671051038255316,
+      "loss": 0.4653009635305819,
       "n_samples": 6783
     },
     {
-      "loss": 0.46794115915756107,
+      "loss": 0.4380375076610923,
       "n_samples": 6783
     },
     {
-      "loss": 0.4293930456997915,
+      "loss": 0.4258159104875806,
       "n_samples": 6783
     },
     {
-      "loss": 0.42624105651716415,
+      "loss": 0.4144523660948226,
       "n_samples": 6783
     },
     {
-      "loss": 0.4131358770446828,
+      "loss": 0.4008358244841981,
       "n_samples": 6783
     },
     {
-      "loss": 0.3946074267790835,
+      "loss": 0.40240038808127093,
       "n_samples": 6783
     },
     {
-      "loss": 0.3898155013755344,
+      "loss": 0.38176763174141226,
       "n_samples": 6783
     },
     {
-      "loss": 0.37861797005733383,
+      "loss": 0.37277743237904,
       "n_samples": 6783
     },
     {
-      "loss": 0.3775682858392304,
+      "loss": 0.3573742728176747,
       "n_samples": 6783
     },
     {
-      "loss": 0.3800349080262064,
+      "loss": 0.3491767022517619,
       "n_samples": 6783
     },
     {
-      "loss": 0.36302345173031675,
+      "loss": 0.3499675623860557,
       "n_samples": 6783
     },
     {
-      "loss": 0.3429561740842766,
+      "loss": 0.35079578643841286,
       "n_samples": 6783
     },
     {
-      "loss": 0.3445638883004898,
+      "loss": 0.3414057292594471,
       "n_samples": 6783
     },
     {
-      "loss": 0.318970229203733,
+      "loss": 0.300529963052257,
       "n_samples": 6783
     },
     {
-      "loss": 0.30179278279904437,
+      "loss": 0.2961940902990875,
       "n_samples": 6783
     },
     {
-      "loss": 0.2887343142006437,
+      "loss": 0.2907693515383844,
       "n_samples": 6783
     },
     {
-      "loss": 0.29240367556855545,
+      "loss": 0.2809350616734551,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.28143580470326973,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.2664423378391215,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.2745858784487654,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.26682337215652197,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.2681302405486289,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.26258669999889017,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.2608744821883436,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.239722755447208,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.24175641130912484,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.23785491213674798,
+      "n_samples": 6783
+    },
+    {
+      "loss": 0.23117999019839675,
       "n_samples": 6783
     }
   ],
   "val": [
     {
-      "loss": 0.7350345371841441,
+      "loss": 0.7379055186813953,
       "n_samples": 2907
     },
     {
-      "loss": 0.7165568811318536,
+      "loss": 0.7269540305477178,
       "n_samples": 2907
     },
     {
-      "loss": 0.7251406249862214,
+      "loss": 0.6927794775152518,
       "n_samples": 2907
     },
     {
-      "loss": 0.6836505264587159,
+      "loss": 0.6652627856533758,
       "n_samples": 2907
     },
     {
-      "loss": 0.6747132955771933,
+      "loss": 0.6721626692594103,
       "n_samples": 2907
     },
     {
-      "loss": 0.6691136244936912,
+      "loss": 0.6660812889787394,
       "n_samples": 2907
     },
     {
-      "loss": 0.6337480902323249,
+      "loss": 0.6329412659009298,
       "n_samples": 2907
     },
     {
-      "loss": 0.6600317959527934,
+      "loss": 0.6395346636332554,
       "n_samples": 2907
     },
     {
-      "loss": 0.6439923948855346,
+      "loss": 0.6228037932749914,
       "n_samples": 2907
     },
     {
-      "loss": 0.643800035575267,
+      "loss": 0.627341245329581,
       "n_samples": 2907
     },
     {
-      "loss": 0.6181512585221839,
+      "loss": 0.6399482499704272,
       "n_samples": 2907
     },
     {
-      "loss": 0.6442458634939151,
+      "loss": 0.6136556721283145,
       "n_samples": 2907
     },
     {
-      "loss": 0.6344759362359862,
+      "loss": 0.6253821217484764,
       "n_samples": 2907
     },
     {
-      "loss": 0.6501405371457472,
+      "loss": 0.6535878175511408,
       "n_samples": 2907
     },
     {
-      "loss": 0.6098835162990152,
+      "loss": 0.6123772015635566,
       "n_samples": 2907
     },
     {
-      "loss": 0.6366627322138894,
+      "loss": 0.648116178003258,
       "n_samples": 2907
     },
     {
-      "loss": 0.6171610150646417,
+      "loss": 0.6141229696318092,
       "n_samples": 2907
     },
     {
-      "loss": 0.6358801012273748,
+      "loss": 0.6307853255273552,
       "n_samples": 2907
     },
     {
-      "loss": 0.6239976831059871,
+      "loss": 0.6422428293329848,
       "n_samples": 2907
     },
     {
-      "loss": 0.6683828232827201,
+      "loss": 0.6552245357949421,
       "n_samples": 2907
     },
     {
-      "loss": 0.6655785786478143,
+      "loss": 0.6614342853503823,
       "n_samples": 2907
     },
     {
-      "loss": 0.6152775046503088,
+      "loss": 0.6153881246378465,
       "n_samples": 2907
     },
     {
-      "loss": 0.6202247662153858,
+      "loss": 0.6298057977526632,
       "n_samples": 2907
     },
     {
-      "loss": 0.648199727435189,
+      "loss": 0.6117005377321011,
       "n_samples": 2907
     },
     {
-      "loss": 0.6473217075085124,
+      "loss": 0.618167482426702,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6103124622220963,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6203263100988888,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.617202088939065,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6263537556599373,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6461186489679168,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6289772454163526,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6347806630652919,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6358688624302136,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.646931814478975,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6245040218978556,
+      "n_samples": 2907
+    },
+    {
+      "loss": 0.6267098000482632,
       "n_samples": 2907
     }
   ]