depth of 1 led to consistently bad learning

newsearch.py

@@ -30,7 +30,7 @@ learning_rate_values = [1e-3]
 alpha_values = [1.0]
 # widths = [2**k for k in range(4, 13)]
 # depths = [1, 2, 4, 8, 16]
-widths, depths = [64, 128, 256], [4, 8]
+widths, depths = [512], [4]
 
 batch_size_values = [256]
 max_epochs_values = [20]
@@ -38,8 +38,8 @@ seeds = list(range(21, 1992))
 optimizers = [
     # "Adagrad",
     "Adam",
-    "SGD",
-    "AdamW",
+    # "SGD",
+    # "AdamW",
     # "LBFGS",
     # "RAdam",
     # "RMSprop",
@@ -63,6 +63,9 @@ all_params = [
 # perform random search with a limit
 search_params = sample(all_params, min(NUM_JOBS, len(all_params)))
 
+# --trainer.callbacks+ lightning.pytorch.callbacks.EarlyStopping \
+# --trainer.callbacks.init_args.monitor hp_metric \
+
 for idx, params in enumerate(search_params):
     a, lr, bs, me, s, w, d, opt = params
     # cmd = f"cd ~/colors && python main.py --alpha {a} --lr {lr} --bs {bs} --max_epochs {me} --seed {s} --width {w}"
@@ -87,19 +90,17 @@ python newmain.py fit \
 --trainer.callbacks.init_args.save_interval 0 \
 --optimizer torch.optim.{opt} \
 --optimizer.init_args.lr {lr} \
---lr_scheduler lightning.pytorch.cli.ReduceLROnPlateau \
---lr_scheduler.init_args.monitor hp_metric \
---lr_scheduler.init_args.factor 0.05 \
---lr_scheduler.init_args.patience 5 \
---lr_scheduler.init_args.cooldown 10 \
---lr_scheduler.init_args.verbose true
+--trainer.callbacks+ lightning.pytorch.callbacks.LearningRateFinder
+# --lr_scheduler lightning.pytorch.cli.ReduceLROnPlateau \
+# --lr_scheduler.init_args.monitor hp_metric \
+# --lr_scheduler.init_args.factor 0.05 \
+# --lr_scheduler.init_args.patience 5 \
+# --lr_scheduler.init_args.cooldown 10 \
+# --lr_scheduler.init_args.verbose true
 """
-    test_cmd = f"{cmd.strip()} --print_config > out/config_v{idx:04d}.txt"
-
     # job_name = f"color2_{bs}_{a}_{lr:2.2e}"
     # job_plugin.run(cmd, machine=Machine.T4, name=job_name)
     print(f"Running {params}: {cmd}")
-    cmd = f"{test_cmd.strip()} && {cmd}"
     try:
         # Run the command and wait for it to complete
         # subprocess.run(test_cmd, shell=True, check=True)