reformat code using black -l 200

yliuhz · Oct 11, 2021 · 61a9cd3 · 61a9cd3
1 parent 1e5ec09
commit 61a9cd3
Show file tree

Hide file tree

Showing 9 changed files with 27 additions and 82 deletions.
diff --git a/datasets/caltech101.py b/datasets/caltech101.py
@@ -28,9 +28,7 @@ def __init__(self, cfg):
         if os.path.exists(self.split_path):
             train, val, test = OxfordPets.read_split(self.split_path, self.image_dir)
         else:
-            train, val, test = DTD.read_and_split_data(
-                self.image_dir, ignored=IGNORED, new_cnames=NEW_CNAMES
-            )
+            train, val, test = DTD.read_and_split_data(self.image_dir, ignored=IGNORED, new_cnames=NEW_CNAMES)
             OxfordPets.save_split(train, val, test, self.split_path, self.image_dir)
 
         num_shots = cfg.DATASET.NUM_SHOTS

diff --git a/datasets/dtd.py b/datasets/dtd.py
@@ -16,9 +16,7 @@ def __init__(self, cfg):
         root = os.path.abspath(os.path.expanduser(cfg.DATASET.ROOT))
         self.dataset_dir = os.path.join(root, self.dataset_dir)
         self.image_dir = os.path.join(self.dataset_dir, "images")
-        self.split_path = os.path.join(
-            self.dataset_dir, "split_zhou_DescribableTextures.json"
-        )
+        self.split_path = os.path.join(self.dataset_dir, "split_zhou_DescribableTextures.json")
 
         if os.path.exists(self.split_path):
             train, val, test = OxfordPets.read_split(self.split_path, self.image_dir)
@@ -33,9 +31,7 @@ def __init__(self, cfg):
         super().__init__(train_x=train, val=val, test=test)
 
     @staticmethod
-    def read_and_split_data(
-        image_dir, p_trn=0.5, p_val=0.2, ignored=[], new_cnames=None
-    ):
+    def read_and_split_data(image_dir, p_trn=0.5, p_val=0.2, ignored=[], new_cnames=None):
         # The data are supposed to be organized into the following structure
         # =============
         # images/
@@ -48,9 +44,7 @@ def read_and_split_data(
         categories.sort()
 
         p_tst = 1 - p_trn - p_val
-        print(
-            f"Splitting into {p_trn:.0%} train, {p_val:.0%} val, and {p_tst:.0%} test"
-        )
+        print(f"Splitting into {p_trn:.0%} train, {p_val:.0%} val, and {p_tst:.0%} test")
 
         def _collate(ims, y, c):
             items = []

diff --git a/datasets/eurosat.py b/datasets/eurosat.py
@@ -33,9 +33,7 @@ def __init__(self, cfg):
         if os.path.exists(self.split_path):
             train, val, test = OxfordPets.read_split(self.split_path, self.image_dir)
         else:
-            train, val, test = DTD.read_and_split_data(
-                self.image_dir, new_cnames=NEW_CNAMES
-            )
+            train, val, test = DTD.read_and_split_data(self.image_dir, new_cnames=NEW_CNAMES)
             OxfordPets.save_split(train, val, test, self.split_path, self.image_dir)
 
         num_shots = cfg.DATASET.NUM_SHOTS
@@ -49,8 +47,6 @@ def update_classname(self, dataset_old):
         for item_old in dataset_old:
             cname_old = item_old.classname
             cname_new = NEW_CLASSNAMES[cname_old]
-            item_new = Datum(
-                impath=item_old.impath, label=item_old.label, classname=cname_new
-            )
+            item_new = Datum(impath=item_old.impath, label=item_old.label, classname=cname_new)
             dataset_new.append(item_new)
         return dataset_new
diff --git a/datasets/oxford_flowers.py b/datasets/oxford_flowers.py
@@ -20,9 +20,7 @@ def __init__(self, cfg):
         self.image_dir = os.path.join(self.dataset_dir, "jpg")
         self.label_file = os.path.join(self.dataset_dir, "imagelabels.mat")
         self.lab2cname_file = os.path.join(self.dataset_dir, "cat_to_name.json")
-        self.split_path = os.path.join(
-            self.dataset_dir, "split_zhou_OxfordFlowers.json"
-        )
+        self.split_path = os.path.join(self.dataset_dir, "split_zhou_OxfordFlowers.json")
 
         if os.path.exists(self.split_path):
             train, val, test = OxfordPets.read_split(self.split_path, self.image_dir)
@@ -50,9 +48,7 @@ def read_data(self):
         def _collate(ims, y, c):
             items = []
             for im in ims:
-                item = Datum(
-                    impath=im, label=y - 1, classname=c  # convert to 0-based label
-                )
+                item = Datum(impath=im, label=y - 1, classname=c)  # convert to 0-based label
                 items.append(item)
             return items
 

diff --git a/datasets/stanford_cars.py b/datasets/stanford_cars.py
@@ -19,9 +19,7 @@ def __init__(self, cfg):
         if os.path.exists(self.split_path):
             train, val, test = OxfordPets.read_split(self.split_path, self.dataset_dir)
         else:
-            trainval_file = os.path.join(
-                self.dataset_dir, "devkit", "cars_train_annos.mat"
-            )
+            trainval_file = os.path.join(self.dataset_dir, "devkit", "cars_train_annos.mat")
             test_file = os.path.join(self.dataset_dir, "cars_test_annos_withlabels.mat")
             meta_file = os.path.join(self.dataset_dir, "devkit", "cars_meta.mat")
             trainval = self.read_data("cars_train", trainval_file, meta_file)

diff --git a/lpclip/feat_extractor.py b/lpclip/feat_extractor.py
@@ -171,25 +171,19 @@ def main(args):
     parser = argparse.ArgumentParser()
     parser.add_argument("--root", type=str, default="", help="path to dataset")
     parser.add_argument("--output-dir", type=str, default="", help="output directory")
-    parser.add_argument(
-        "--config-file", type=str, default="", help="path to config file"
-    )
+    parser.add_argument("--config-file", type=str, default="", help="path to config file")
     parser.add_argument(
         "--dataset-config-file",
         type=str,
         default="",
         help="path to config file for dataset setup",
     )
     parser.add_argument("--num-shot", type=int, default=1, help="number of shots")
-    parser.add_argument(
-        "--split", type=str, choices=["train", "val", "test"], help="which split"
-    )
+    parser.add_argument("--split", type=str, choices=["train", "val", "test"], help="which split")
     parser.add_argument("--trainer", type=str, default="", help="name of trainer")
     parser.add_argument("--backbone", type=str, default="", help="name of CNN backbone")
     parser.add_argument("--head", type=str, default="", help="name of head")
-    parser.add_argument(
-        "--seed", type=int, default=-1, help="only positive value enables a fixed seed"
-    )
+    parser.add_argument("--seed", type=int, default=-1, help="only positive value enables a fixed seed")
     parser.add_argument("--eval-only", action="store_true", help="evaluation only")
     args = parser.parse_args()
     main(args)
diff --git a/lpclip/linear_probe.py b/lpclip/linear_probe.py
@@ -7,9 +7,7 @@
 parser.add_argument("--dataset", type=str, default="", help="path to dataset")
 parser.add_argument("--num_step", type=int, default=8, help="number of steps")
 parser.add_argument("--num_run", type=int, default=10, help="number of runs")
-parser.add_argument(
-    "--feature_dir", type=str, default="clip_feat", help="feature dir path"
-)
+parser.add_argument("--feature_dir", type=str, default="clip_feat", help="feature dir path")
 args = parser.parse_args()
 
 dataset = args.dataset
@@ -30,17 +28,13 @@
     test_acc_step_list = np.zeros([args.num_run, args.num_step])
     for seed in range(1, args.num_run + 1):
         np.random.seed(seed)
-        print(
-            f"-- Seed: {seed} --------------------------------------------------------------"
-        )
+        print(f"-- Seed: {seed} --------------------------------------------------------------")
         # Sampling
         all_label_list = np.unique(train_label)
         selected_idx_list = []
         for label in all_label_list:
             label_collection = np.where(train_label == label)[0]
-            selected_idx = np.random.choice(
-                label_collection, size=num_shot, replace=False
-            )
+            selected_idx = np.random.choice(label_collection, size=num_shot, replace=False)
             selected_idx_list.extend(selected_idx)
 
         fewshot_train_feature = train_feature[selected_idx_list]
@@ -50,9 +44,7 @@
         val_selected_idx_list = []
         for label in all_label_list:
             label_collection = np.where(val_label == label)[0]
-            selected_idx = np.random.choice(
-                label_collection, size=val_num_shot, replace=False
-            )
+            selected_idx = np.random.choice(label_collection, size=val_num_shot, replace=False)
             val_selected_idx_list.extend(selected_idx)
 
         fewshot_val_feature = val_feature[val_selected_idx_list]
@@ -62,9 +54,7 @@
         search_list = [1e6, 1e4, 1e2, 1, 1e-2, 1e-4, 1e-6]
         acc_list = []
         for c_weight in search_list:
-            clf = LogisticRegression(
-                solver="lbfgs", max_iter=1000, penalty="l2", C=c_weight
-            ).fit(fewshot_train_feature, fewshot_train_label)
+            clf = LogisticRegression(solver="lbfgs", max_iter=1000, penalty="l2", C=c_weight).fit(fewshot_train_feature, fewshot_train_label)
             pred = clf.predict(fewshot_val_feature)
             acc_val = sum(pred == fewshot_val_label) / len(fewshot_val_label)
             acc_list.append(acc_val)
@@ -77,16 +67,12 @@
         c_left, c_right = 1e-1 * c_peak, 1e1 * c_peak
 
         def binary_search(c_left, c_right, seed, step, test_acc_step_list):
-            clf_left = LogisticRegression(
-                solver="lbfgs", max_iter=1000, penalty="l2", C=c_left
-            ).fit(fewshot_train_feature, fewshot_train_label)
+            clf_left = LogisticRegression(solver="lbfgs", max_iter=1000, penalty="l2", C=c_left).fit(fewshot_train_feature, fewshot_train_label)
             pred_left = clf_left.predict(fewshot_val_feature)
             acc_left = sum(pred_left == fewshot_val_label) / len(fewshot_val_label)
             print("Val accuracy (Left): {:.2f}".format(100 * acc_left), flush=True)
 
-            clf_right = LogisticRegression(
-                solver="lbfgs", max_iter=1000, penalty="l2", C=c_right
-            ).fit(fewshot_train_feature, fewshot_train_label)
+            clf_right = LogisticRegression(solver="lbfgs", max_iter=1000, penalty="l2", C=c_right).fit(fewshot_train_feature, fewshot_train_label)
             pred_right = clf_right.predict(fewshot_val_feature)
             acc_right = sum(pred_right == fewshot_val_label) / len(fewshot_val_label)
             print("Val accuracy (Right): {:.2f}".format(100 * acc_right), flush=True)
@@ -110,13 +96,9 @@ def binary_search(c_left, c_right, seed, step, test_acc_step_list):
             print("Test Accuracy: {:.2f}".format(test_acc), flush=True)
             test_acc_step_list[seed - 1, step] = test_acc
 
-            saveline = "{}, seed {}, {} shot, weight {}, test_acc {:.2f}\n".format(
-                dataset, seed, num_shot, c_final, test_acc
-            )
+            saveline = "{}, seed {}, {} shot, weight {}, test_acc {:.2f}\n".format(dataset, seed, num_shot, c_final, test_acc)
             with open(
-                "./report/{}_s{}r{}_details.txt".format(
-                    args.feature_dir, args.num_step, args.num_run
-                ),
+                "./report/{}_s{}r{}_details.txt".format(args.feature_dir, args.num_step, args.num_run),
                 "a+",
             ) as writer:
                 writer.write(saveline)
@@ -133,16 +115,12 @@ def binary_search(c_left, c_right, seed, step, test_acc_step_list):
                 f"{dataset}, {num_shot} Shot, Round {step}: {c_left}/{c_right}",
                 flush=True,
             )
-            c_left, c_right, seed, step, test_acc_step_list = binary_search(
-                c_left, c_right, seed, step, test_acc_step_list
-            )
+            c_left, c_right, seed, step, test_acc_step_list = binary_search(c_left, c_right, seed, step, test_acc_step_list)
     # save results of last step
     test_acc_list = test_acc_step_list[:, -1]
     acc_mean = np.mean(test_acc_list)
     acc_std = np.std(test_acc_list)
-    save_line = "{}, {} Shot, Test acc stat: {:.2f} ({:.2f})\n".format(
-        dataset, num_shot, acc_mean, acc_std
-    )
+    save_line = "{}, {} Shot, Test acc stat: {:.2f} ({:.2f})\n".format(dataset, num_shot, acc_mean, acc_std)
     print(save_line, flush=True)
     with open(
         "./report/{}_s{}r{}.txt".format(args.feature_dir, args.num_step, args.num_run),

diff --git a/trainers/coop.py b/trainers/coop.py
@@ -66,9 +66,7 @@ def __init__(self, cfg, classnames, clip_model):
         ctx_dim = clip_model.ln_final.weight.shape[0]
         clip_imsize = clip_model.visual.input_resolution
         cfg_imsize = cfg.INPUT.SIZE[0]
-        assert (
-            cfg_imsize == clip_imsize
-        ), f"cfg_imsize ({cfg_imsize}) must equal to clip_imsize ({clip_imsize})"
+        assert cfg_imsize == clip_imsize, f"cfg_imsize ({cfg_imsize}) must equal to clip_imsize ({clip_imsize})"
 
         if ctx_init:
             # use given words to initialize context vectors
@@ -242,9 +240,7 @@ def build_model(self):
         # NOTE: only give prompt_learner to the optimizer
         self.optim = build_optimizer(self.model.prompt_learner, cfg.OPTIM)
         self.sched = build_lr_scheduler(self.optim, cfg.OPTIM)
-        self.register_model(
-            "prompt_learner", self.model.prompt_learner, self.optim, self.sched
-        )
+        self.register_model("prompt_learner", self.model.prompt_learner, self.optim, self.sched)
 
         device_count = torch.cuda.device_count()
         if device_count > 1:
@@ -304,9 +300,6 @@ def load_model(self, directory, epoch=None):
             if "token_suffix" in state_dict:
                 del state_dict["token_suffix"]
 
-            print(
-                "Loading weights to {} "
-                'from "{}" (epoch = {})'.format(name, model_path, epoch)
-            )
+            print("Loading weights to {} " 'from "{}" (epoch = {})'.format(name, model_path, epoch))
             # set strict=False
             self._models[name].load_state_dict(state_dict, strict=False)
diff --git a/trainers/zsclip.py b/trainers/zsclip.py
@@ -93,9 +93,7 @@ def build_model(self):
             text_features = text_features / text_features.norm(dim=-1, keepdim=True)
             mean_text_features = mean_text_features + text_features
         mean_text_features = mean_text_features / num_temp
-        mean_text_features = mean_text_features / mean_text_features.norm(
-            dim=-1, keepdim=True
-        )
+        mean_text_features = mean_text_features / mean_text_features.norm(dim=-1, keepdim=True)
 
         self.text_features = mean_text_features
         self.clip_model = clip_model