update and fix

.github/workflows/exe_build_action.yml

@@ -19,8 +19,8 @@ jobs:
           spec: 'src/comsol/cmdline.py'
           requirements: 'requirements.txt'
           exe_path: 'dist/windows'
-          upload_exe_with_name: 'consol_CLI'
-          options: --onefile, --name "consol_CLI"
+          upload_exe_with_name: 'comsol_CLI'
+          options: --onefile, --name "comsol_CLI"
       - name: create release
         id: create_release
         uses: ncipollo/[email protected]
@@ -40,5 +40,5 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         with:
           files: |
-            dist/windows/consol_CLI.exe
+            dist/windows/comsol_CLI.exe
             dist/windows/README.md

config/cell.yaml

@@ -16,6 +16,6 @@ cell:
     step: 0.001
 
 train:
-  epoch: 500
+  epoch: 200
   batch_size: 16
   lr: 0.001

config/cell_local.yaml

@@ -0,0 +1,21 @@
+cell:
+  zeta:
+    init: 0.0005
+    max: 180
+    min: 0
+    step: 30
+  rr:
+    init: 0.001
+    max: 0.0045
+    min: 0.0035
+    step: 0.0005
+  rrr:
+    init: 0.001
+    max: 0.003
+    min: 0.001
+    step: 0.001
+
+train:
+  epoch: 300
+  batch_size: 8
+  lr: 0.001

src/comsol/cmdline.py

@@ -44,7 +44,10 @@ def train(saved, config, ckpt_path):
     dataset = BandDataset(saved)
     model = MLP()
     trainer = Trainer(dataset, model, cfg)
-    trainer.train()
+    try:
+        trainer.train()
+    except KeyboardInterrupt:
+        trainer.save_ckpt(f"earlystop_best_{trainer.best_loss:.3f}.pth", best=True)
 
 
 @main.command()

src/comsol/ga.py

@@ -1,8 +1,8 @@
 import numpy
 import pygad
 import torch
-
 from loguru import logger
+
 from comsol.model import MLP
 from comsol.utils import BandDataset
 
@@ -19,7 +19,7 @@ def fitness_func(ga_instance, solution, solution_idx):
         Bs: numpy.ndarray = (
             net(torch.tensor([solution]).float()).detach().numpy().flatten()
         )
-        return abs(Bs[2] - Bs[0])
+        return abs((Bs[1] + Bs[2]) / 2 - Bs[0])
 
     fitness_function = fitness_func
 
@@ -57,6 +57,10 @@ def fitness_func(ga_instance, solution, solution_idx):
 
     prediction = net(torch.tensor([solution]).float()).detach().numpy().flatten()
     solution, prediction = dataset.denormalization(solution, prediction)
+    solution[0] = solution[0] * 360
     logger.info(f"Parameters of the best solution : {solution}")
     logger.info(f"Predicted output based on the best solution : {prediction}")
-    logger.info(f"Fitness value of the best solution = {solution_fitness}")
+    logger.info(f"Fitness value of the best solution : {solution_fitness}")
+    logger.info(
+        f"Fitness value of the best solution denormalization= {solution_fitness * (dataset.res_max - dataset.res_min) + dataset.res_min:.5e}"
+    )

src/comsol/utils.py

@@ -1,3 +1,4 @@
+import functools
 import pickle
 from dataclasses import dataclass
 from datetime import datetime
@@ -124,7 +125,7 @@ def train(self):
             self.test()
         self.save_ckpt("lastest")
         logger.info(f"Training finished, best loss: {self.best_loss:.3f}")
-        self.save_ckpt(f"best_loss_{self.best_loss:.3f}")
+        self.save_ckpt(f"best_loss_{self.best_loss:.3f}", best=True)
 
     def test(self):
         self.model.eval()
@@ -142,7 +143,7 @@ def test(self):
             self.best_loss = now_loss
             self.best_ckpt = self.model.state_dict()
 
-    def save_ckpt(self, name):
+    def save_ckpt(self, name, best=False):
         ckpt_path = Path(f"ckpt") / f"{self.start_time:%Y.%m.%d_%H.%M.%S}"
         ckpt_path.mkdir(parents=True, exist_ok=True)
 
@@ -151,7 +152,10 @@ def save_ckpt(self, name):
         if not cfg_path.exists():
             self.cfg.dump(cfg_path)
             logger.info(f"Dumped config to {cfg_path}")
-        torch.save(self.model.state_dict(), pth_path)
+        if best:
+            torch.save(self.best_ckpt, pth_path)
+        else:
+            torch.save(self.model.state_dict(), pth_path)
         logger.info(f"Saved model to {pth_path}")