Fix circleci

.circleci/config.yml

@@ -6,7 +6,7 @@ orbs:
 executors:
   python-executor:
     docker:
-      - image: cimg/python:3.10.3
+      - image: cimg/python:3.11
     working_directory: ~/repo
 
 
@@ -254,11 +254,13 @@ jobs:
           command: |
             echo -e 'GRU Vanilla test'
             coverage run flood_forecast/trainer.py -p tests/gru_vanilla.json
+            echo -e 'Running classification test'
             coverage run flood_forecast/trainer.py -p tests/classification_test.json
+            echo -e 'Running informer single test'
             coverage run flood_forecast/trainer.py -p tests/test_inf_single.json
-            echo -e 'test informer single target'
-            echo -e 'test multi informer'
+            echo -e 'Running transformer_gaussian test'
             coverage run flood_forecast/trainer.py -p tests/transformer_gaussian.json
+            echo -e 'Running multi_decoder_test.json'
             coverage run flood_forecast/trainer.py -p tests/multi_decoder_test.json
             echo -e 'training multi-task-decoder'
             coverage run flood_forecast/trainer.py -p tests/multitask_decoder.json
@@ -274,14 +276,7 @@ jobs:
             coverage run flood_forecast/trainer.py -p tests/probabilistic_linear_regression_test.json
             echo -e 'running meta-data fusion'
             coverage run flood_forecast/trainer.py -p tests/meta_data_test.json
-            echo -e 'running da_rnn_test \n'
-            coverage run flood_forecast/trainer.py -p tests/da_rnn.json
-            echo -e 'running auto test \n'
-            coverage run flood_forecast/meta_train.py -p tests/auto_encoder.json
-            echo -e 'running trainer_lstm_test \n '
-            coverage run flood_forecast/trainer.py -p tests/lstm_test.json
-            echo -e 'running trainer_multi_test \n'
-            coverage run flood_forecast/trainer.py -p tests/multi_test.json
+
       - store_test_results:
           path: test-results
 
@@ -319,23 +314,32 @@ jobs:
           name: Trainer1 tests
           when: always
           command: |
+            echo -e 'running test_informer \n'
             coverage run flood_forecast/trainer.py -p tests/test_informer.json
+            echo -e 'running test_iTransformer \n'
             coverage run flood_forecast/trainer.py -p tests/test_iTransformer.json
+            echo -e 'running tsmixer_test \n'
             coverage run flood_forecast/trainer.py -p tests/tsmixer_test.json
+            echo -e 'running transformer_b_series.json \n'
             coverage run flood_forecast/trainer.py -p tests/transformer_b_series.json
+            echo -e 'running crossfroer.json \n'
             coverage run flood_forecast/trainer.py -p tests/cross_former.json
+            echo -e 'running nlinear.json \n'
             coverage run flood_forecast/trainer.py -p tests/nlinear.json
+            echo -e 'running dsanet_3\n'
             coverage run flood_forecast/trainer.py -p tests/dsanet_3.json
+            echo -e 'running variable_autoencoderl.json\n'
             coverage run flood_forecast/trainer.py -p tests/variable_autoencoderl.json
+            echo -e 'running dlinear\n'
             coverage run flood_forecast/trainer.py -p tests/dlinear.json
+            echo -e 'running custom_encode.json\n'
             coverage run flood_forecast/trainer.py -p tests/custom_encode.json
+            echo -e 'running multi_decoder_test \n'
             coverage run flood_forecast/trainer.py -p tests/multi_decoder_test.json
+            echo -e 'test_dual \n'
             coverage run flood_forecast/trainer.py -p tests/test_dual.json
+            echo -e 'running dsanet \n'
             coverage run flood_forecast/trainer.py -p tests/dsanet.json
-            echo -e 'running trainer_decoder_test \n'
-            coverage run flood_forecast/trainer.py -p tests/decoder_test.json
-            echo -e 'running trainer_full_transformer_test'
-            coverage run flood_forecast/trainer.py -p  tests/full_transformer.json
 
       - store_test_results:
           path: test-results
@@ -349,6 +353,54 @@ jobs:
           when: always
           command: bash <(curl -s https://codecov.io/bash) -cF python
 
+  trainer_test2:
+      <<: *defaults
+      steps:
+        - attach_workspace:
+            at: ~/repo
+        - restore_cache:  # ensure this step occurs *before* installing dependencies
+            key: requirements-v2-{{ .Branch }}-{{ checksum "requirements.txt" }}
+
+        - run:
+            name: install dependencies
+            command: |
+              pip install pipenv
+              pip install --force-reinstall --upgrade --quiet -r requirements.txt
+              pip install coverage
+              python setup.py develop --no-deps
+        - save_cache:
+            key: requirements-v2-{{ .Branch }}-{{ checksum "requirements.txt" }}
+            paths:
+              - ~/.local
+
+        - run:
+            name: Trainer tests 2
+            when: always
+            command: |
+              echo -e 'running trainer_lstm_test \n '
+              coverage run flood_forecast/trainer.py -p tests/lstm_test.json
+              echo -e 'running trainer_multi_test \n'
+              coverage run flood_forecast/trainer.py -p tests/multi_test.json
+              echo -e 'running trainer_full_transformer_test \n'
+              coverage run flood_forecast/trainer.py -p  tests/full_transformer.json
+              echo -e 'running da_rnn_test \n'
+              coverage run flood_forecast/trainer.py -p tests/da_rnn.json
+              echo -e 'running auto_encoder \n'
+              coverage run flood_forecast/meta_train.py -p tests/auto_encoder.json
+              echo -e 'running trainer_decoder_test   \n'
+              coverage run flood_forecast/trainer.py -p tests/decoder_test.json
+        - store_test_results:
+            path: test-results
+
+        - store_artifacts:
+            path: test-results
+            destination: test-results-trainer
+
+        - run:
+            name: upload results
+            when: always
+            command: bash <(curl -s https://codecov.io/bash) -cF python
+
   plot_test:
     <<: *defaults
     steps:
@@ -420,3 +472,7 @@ workflows:
       - trainer_test1:
           requires:
             - setup_and_install
+      - trainer_test2:
+          requires:
+            - setup_and_install
+

docs/source/basic_ae.rst

@@ -4,4 +4,4 @@ Simple AE
 .. automodule:: flood_forecast.meta_models.basic_ae
     :members:
 
-A simple auto-encoder
+A simple auto-encoder model.

docs/source/basic_utils.rst

@@ -1,5 +1,7 @@
-Basic GCP Utils
+Basic Google Cloud Platform Utilities
 ================
 
+Flow Forecast natively integrates with Google Cloud Platform.
+
 .. automodule:: flood_forecast.gcp_integration.basic_utils
     :members:

docs/source/crossformer.rst

@@ -0,0 +1,4 @@
+Crossformer
+=========================
+.. automodule:: flood_forecast.transformer_xl.crossformer
+    :members: 

docs/source/custom_opt.rst

@@ -1,6 +1,5 @@
-Custom Optimizations
+Custom Optimizers and more
 ====================
 
 .. automodule:: flood_forecast.custom.custom_opt
     :members:
-sss

docs/source/d_linear.rst

@@ -0,0 +1,5 @@
+D and N Linear
+==================
+
+.. automodule:: flood_forecast.basic.d_n_linear
+    :members:

docs/source/index.rst

@@ -64,6 +64,7 @@ of datasets. This documentation describes the internal Python code that makes up
   :caption: TransformerXL:
 
   dummy_torch
+  itransformer
   lower_upper_config
   multi_head_base
   transformer_basic

docs/source/itransformer.rst

@@ -0,0 +1,5 @@
+I-Transformer Model.
+==================
+
+.. automodule:: flood_forecast.transformer_xl.itransformer
+    :members:

flood_forecast/evaluator.py

@@ -526,7 +526,6 @@ def generate_predictions(
         history_dim = history
     else:
         history_dim = history.unsqueeze(0).to(model.device)
-    print("Add debugging crap below")
     if decoder_params is None:
         end_tensor = generate_predictions_non_decoded(
             model, df, test_data, history_dim, forecast_length, hours_to_forecast,
@@ -615,8 +614,6 @@ def generate_predictions_non_decoded(
         end_tensor = torch.cat(all_tensor, axis=0).to("cpu").detach()[:-remainder]
     else:
         end_tensor = torch.cat(all_tensor, axis=0).to("cpu").detach()
-
-    print(end_tensor.shape)  # Dimension now is (n_time_steps_to_forecast_steps)!! i.e [16]
     return end_tensor
 
 

flood_forecast/model_dict_function.py

@@ -27,7 +27,7 @@
 
 
 """
-Utility dictionaries to map a string to a c class
+Utility dictionaries to map a string to a class
 """
 pytorch_model_dict = {
     "MultiAttnHeadSimple": MultiAttnHeadSimple,

flood_forecast/preprocessing/pytorch_loaders.py

@@ -197,7 +197,8 @@ def __init__(self, series_id_col: str, main_params: dict, return_method: str, re
             print(new_df.columns)
         self.listed_vals = df_list
         self.__make_unique_dict__()
-        self.__validate_data__in_df()
+        if return_all:
+            self.__validate_data__in_df()
         print(self.unique_dict)
         print("unique dict")
 
@@ -644,7 +645,7 @@ def get_item_auto_encoder(self, idx):
             return the_seq.float(), the_seq.float()
 
     def pad_input_data(self, sequence: int):
-        """Pads a sequence to a specified length
+        """Pads a sequence to a specified length.
         """
         if self.pad_length > sequence.shape[0]:
             pad_dim = self.pad_length - sequence.shape[0]
@@ -657,13 +658,6 @@ def __getitem__(self, idx: int):
         return tasks[self.task](idx)
 
 
-class CSVResultsHolder(object):
-    def __init__(self, historical_rows, all_rows_orig, targ_idx) -> None:
-        self.historical_rows = historical_rows
-        self.all_rows_orig = all_rows_orig
-        self.targ_idx = targ_idx
-
-
 class SeriesIDTestLoader(CSVSeriesIDLoader):
     def __init__(self, series_id_col: str, main_params: dict, return_method: str, forecast_total=336, return_all=True):
         """_summary_
@@ -672,11 +666,11 @@ def __init__(self, series_id_col: str, main_params: dict, return_method: str, fo
         :type series_id_col: str
         :param main_params: The core params used to instantiate the CSVSeriesIDLoader
         :type main_params: dict
-        :param return_method: _description_D
+        :param return_method: The method of return
         :type return_method: str
         :param return_all: _description_, defaults to True
         :type return_all: bool, optional
-        :param forecast_total: _description_, defaults to 336
+        :param forecast_total: The total length to forecast, defaults to 336
         :type forecast_total: int, optional
         """
         super().__init__(series_id_col, main_params, return_method, return_all)

flood_forecast/preprocessing/temporal_feats.py

@@ -14,7 +14,7 @@ def create_feature(key: str, value: str, df: pd.DataFrame, dt_column: str):
     :type df: pd.DataFrame
     :param dt_column: The name of the datetime column
     :type dt_column: str
-    :return: The dataframe with the newly added column.
+    :return: The dataframe with the newly added columns.
     :rtype: pd.DataFrame
     """
     if key == "day_of_week":

flood_forecast/transformer_xl/basis_former.py

@@ -0,0 +1 @@
+# TO-DO implement basis former

flood_forecast/transformer_xl/cross_former.py

@@ -33,21 +33,21 @@ def __init__(
         :type forecast_length: int
         :param seg_len: Parameter specific to Crossformer, forecast_history must be divisible by seg_len
         :type seg_len: int
-        :param win_size: _description_, defaults to 4
+        :param win_size: The window size for the segment merge mechanism, defaults to 4 (original paper used 2)
         :type win_size: int, optional
         :param factor: _description_, defaults to 10
         :type factor: int, optional
         :param d_model: _description_, defaults to 512
         :type d_model: int, optional
         :param d_ff: _description_, defaults to 1024
         :type d_ff: int, optional
-        :param n_heads: _description_, defaults to 8
+        :param n_heads: The number of heads in the multi-head attention mechanism, defaults to 8
         :type n_heads: int, optional
-        :param e_layers: _description_, defaults to 3
+        :param e_layers: The number of encoder layers, defaults to 3
         :type e_layers: int, optional
-        :param dropout: _description_, defaults to 0.0
+        :param dropout: The amount of dropout to use when training the model, defaults to 0.0
         :type dropout: float, optional
-        :param baseline: _description_, defaults to False
+        :param baseline: A boolean of whether to use mean of the past time series , defaults to False
         :type baseline: bool, optional
         :param device: _description_, defaults to torch.device("cuda:0")
         :type device: str, optional

flood_forecast/transformer_xl/informer.py

@@ -124,10 +124,7 @@ def forward(self, x_enc: torch.Tensor, x_mark_enc, x_dec, x_mark_dec,
         """
         enc_out = self.enc_embedding(x_enc, x_mark_enc)
         enc_out, _ = self.encoder(enc_out, attn_mask=enc_self_mask)
-        print(type(enc_out))
-        print("thing above")
         dec_out = self.dec_embedding(x_dec, x_mark_dec)
-        print(type(dec_out))
         dec_out = self.decoder(dec_out, enc_out, x_mask=dec_self_mask, cross_mask=dec_enc_mask)
         dec_out = self.projection(dec_out)
         # dec_out = self.end_conv1(dec_out)

flood_forecast/transformer_xl/transformer_bottleneck.py

@@ -268,10 +268,6 @@ def forward(self, series_id: int, x: torch.Tensor):
             embedding_sum = torch.zeros(batch_size, length)
             embedding_sum = embedding_sum.fill_(series_id).type(torch.LongTensor).to(self.device)
             embedding_sum = self.id_embed(embedding_sum)
-        print("shape below")
-        print(embedding_sum.shape)
-        print(x.shape)
-        print(series_id)
         position = torch.tensor(torch.arange(length), dtype=torch.long).to(self.device)
         po_embedding = self.po_embed(position)
         embedding_sum[:] = po_embedding

tests/custom_encode.json

@@ -37,7 +37,7 @@
 
        },
        "lr": 0.3,
-       "epochs": 30,
+       "epochs": 2,
        "batch_size":4
 
     },