Add tests, server, resume training, and project cleanup

- Add 57 unit tests covering generators, models, and pipeline components
- Implement FastAPI HTTP service (server.py) with POST /solve and GET /health
- Add checkpoint resume (断点续训) to both CTC and regression training utils
- Fix device mismatch bug in CTC training (targets/input_lengths on GPU)
- Add pytest dev dependency to pyproject.toml
- Update .gitignore with data/solver/, data/real/, *.log
- Remove PyCharm template main.py
- Update training/__init__.py docs for solver training scripts

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

training/__init__.py

@@ -10,4 +10,6 @@
 - train_3d_rotate.py:        训练 3D 旋转回归 (RegressionCNN)
 - train_3d_slider.py:        训练 3D 滑块回归 (RegressionCNN)
 - train_classifier.py:       训练调度分类器 (CaptchaClassifier)
+- train_slide.py:            训练滑块缺口检测 (GapDetectorCNN)
+- train_rotate_solver.py:    训练旋转角度回归 (RotationRegressor)
 """

training/train_regression_utils.py

@@ -175,10 +175,26 @@ def train_regression_model(
 
     best_mae = float("inf")
     best_tol_acc = 0.0
+    start_epoch = 1
     ckpt_path = CHECKPOINTS_DIR / f"{model_name}.pth"
 
+    # ---- 3.5 断点续训 ----
+    if ckpt_path.exists():
+        ckpt = torch.load(ckpt_path, map_location=device, weights_only=True)
+        model.load_state_dict(ckpt["model_state_dict"])
+        best_tol_acc = ckpt.get("best_tol_acc", 0.0)
+        best_mae = ckpt.get("best_mae", float("inf"))
+        start_epoch = ckpt.get("epoch", 0) + 1
+        # 快进 scheduler 到对应 epoch
+        for _ in range(start_epoch - 1):
+            scheduler.step()
+        print(
+            f"[续训] 从 epoch {start_epoch} 继续, "
+            f"best_tol_acc={best_tol_acc:.4f}, best_mae={best_mae:.2f}"
+        )
+
     # ---- 4. 训练循环 ----
-    for epoch in range(1, cfg["epochs"] + 1):
+    for epoch in range(start_epoch, cfg["epochs"] + 1):
         model.train()
         total_loss = 0.0
         num_batches = 0

training/train_utils.py

@@ -170,10 +170,25 @@ def train_ctc_model(
     ctc_loss = nn.CTCLoss(blank=0, zero_infinity=True)
 
     best_acc = 0.0
+    start_epoch = 1
     ckpt_path = CHECKPOINTS_DIR / f"{model_name}.pth"
 
+    # ---- 3.5 断点续训 ----
+    if ckpt_path.exists():
+        ckpt = torch.load(ckpt_path, map_location=device, weights_only=True)
+        model.load_state_dict(ckpt["model_state_dict"])
+        best_acc = ckpt.get("best_acc", 0.0)
+        start_epoch = ckpt.get("epoch", 0) + 1
+        # 快进 scheduler 到对应 epoch
+        for _ in range(start_epoch - 1):
+            scheduler.step()
+        print(
+            f"[续训] 从 epoch {start_epoch} 继续, "
+            f"best_acc={best_acc:.4f}"
+        )
+
     # ---- 4. 训练循环 ----
-    for epoch in range(1, cfg["epochs"] + 1):
+    for epoch in range(start_epoch, cfg["epochs"] + 1):
         model.train()
         total_loss = 0.0
         num_batches = 0
@@ -181,11 +196,12 @@ def train_ctc_model(
         pbar = tqdm(train_loader, desc=f"Epoch {epoch}/{cfg['epochs']}", leave=False)
         for images, targets, target_lengths, _ in pbar:
             images = images.to(device)
+            targets = targets.to(device)
+            target_lengths = target_lengths.to(device)
 
             logits = model(images)  # (T, B, C)
             T, B, C = logits.shape
-            # cuDNN CTC requires targets/lengths on CPU
-            input_lengths = torch.full((B,), T, dtype=torch.int32)
+            input_lengths = torch.full((B,), T, dtype=torch.int32, device=device)
 
             log_probs = logits.log_softmax(2)
             loss = ctc_loss(log_probs, targets, input_lengths, target_lengths)