feat(pt): allow --init-frz-model for pt model converted from tf model (#5091)

- Add try-catch block for loading frozen models
- Use strict=False when state_dict keys don't match
- Log warnings for model state_dict mismatches
- Prevent crashes when loading models with different architectures"

Fix bug in #5090 

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Bug Fixes**
* Frozen model loading now uses non‑strict mode; missing or unexpected
parameters are logged as warnings instead of causing failure.

* **Tests**
* Added an end-to-end test covering conversion and loading of frozen
models from external formats.
* Test I/O standardized to UTF‑8; teardown extended to remove generated
artifacts.

* **New Files**
* Added a JSON test model configuration for validation and training
workflows.

<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: ChiahsinChu

deepmd/pt/train/training.py

@@ -613,7 +613,13 @@ def single_model_finetune(
 
         if init_frz_model is not None:
             frz_model = torch.jit.load(init_frz_model, map_location=DEVICE)
-            self.model.load_state_dict(frz_model.state_dict())
+            state = frz_model.state_dict()
+            missing, unexpected = self.model.load_state_dict(state, strict=False)
+            if missing or unexpected:
+                log.warning(
+                    "Checkpoint loaded non-strictly. "
+                    f"Missing keys: {missing}, Unexpected keys: {unexpected}"
+                )
 
         # Get model prob for multi-task
         if self.multi_task:

source/tests/pt/model/models/se_e2_a.json

@@ -0,0 +1,77 @@
+{
+  "model": {
+    "type_map": [
+      "O",
+      "H"
+    ],
+    "descriptor": {
+      "type": "se_e2_a",
+      "sel": [
+        23,
+        46
+      ],
+      "rcut_smth": 0.50,
+      "rcut": 4.00,
+      "neuron": [
+        2,
+        4
+      ],
+      "resnet_dt": false,
+      "axis_neuron": 4,
+      "type_one_side": true,
+      "seed": 1,
+      "_comment": " that's all"
+    },
+    "fitting_net": {
+      "neuron": [
+        120
+      ],
+      "resnet_dt": true,
+      "seed": 1,
+      "_comment": " that's all"
+    },
+    "data_stat_nbatch": 20,
+    "_comment": " that's all"
+  },
+  "learning_rate": {
+    "type": "exp",
+    "decay_steps": 5000,
+    "start_lr": 0.001,
+    "stop_lr": 3.51e-8,
+    "_comment": "that's all"
+  },
+  "loss": {
+    "type": "ener",
+    "start_pref_e": 0.02,
+    "limit_pref_e": 1,
+    "start_pref_f": 1000,
+    "limit_pref_f": 1,
+    "_comment": " that's all"
+  },
+  "training": {
+    "training_data": {
+      "systems": [
+        "../data/data_0",
+        "../data/data_1",
+        "../data/data_2"
+      ],
+      "batch_size": 1,
+      "_comment": "that's all"
+    },
+    "validation_data": {
+      "systems": [
+        "../data/data_3"
+      ],
+      "batch_size": 1,
+      "numb_btch": 3,
+      "_comment": "that's all"
+    },
+    "numb_steps": 100000,
+    "seed": 10,
+    "disp_file": "lcurve.out",
+    "disp_freq": 100,
+    "save_freq": 10000,
+    "_comment": "that's all"
+  },
+  "_comment": "that's all"
+}