Module 3

module-3/classic-example/Makefile

@@ -29,12 +29,12 @@ test_all:
 	pytest --cov=classic_example tests/
 
 train_example:
-	python classic_example/cli.py load-cola-data ./data
+	python classic_example/cli.py load-sst2-data ./data
 	python classic_example/cli.py train ./conf/example.json
 	python classic_example/cli.py upload-to-registry example_model /tmp/results
 
 train_fast_ci:
-	python classic_example/cli.py load-cola-data ./data
+	python classic_example/cli.py load-sst2-data ./data
 	python classic_example/cli.py train ./conf/fast.json
 	python classic_example/cli.py upload-to-registry fast-model /tmp/results
 

module-3/classic-example/classic_example/cli.py

@@ -1,14 +1,14 @@
 import typer
 
-from classic_example.data import load_cola_data, load_cola_data_file_input
+from classic_example.data import load_sst2_data, load_sst2_data_file_input
 from classic_example.train import train
 from classic_example.utils import load_from_registry, upload_to_registry
 from classic_example.predictor import run_inference_on_dataframe
 
 app = typer.Typer()
 app.command()(train)
-app.command()(load_cola_data)
-app.command()(load_cola_data_file_input)
+app.command()(load_sst2_data)
+app.command()(load_sst2_data_file_input)
 app.command()(upload_to_registry)
 app.command()(load_from_registry)
 app.command()(run_inference_on_dataframe)

module-3/classic-example/classic_example/data.py

@@ -5,7 +5,7 @@
 from sklearn.model_selection import train_test_split
 
 
-def _get_cola_data(random_state: int = 42):
+def _get_sst2_data(random_state: int = 42):
     dataset = load_dataset("glue", "sst2")
     df_all = ArrowReader.read_table(
         dataset.cache_files["train"][0]["filename"]
@@ -19,24 +19,24 @@ def _get_cola_data(random_state: int = 42):
     return df_train, df_val, df_test
 
 
-def load_cola_data(path_to_save: Path):
+def load_sst2_data(path_to_save: Path):
     path_to_save.mkdir(parents=True, exist_ok=True)
 
-    df_train, df_val, df_test = _get_cola_data()
+    df_train, df_val, df_test = _get_sst2_data()
 
     df_train.to_csv(path_to_save / "train.csv", index=False)
     df_val.to_csv(path_to_save / "val.csv", index=False)
     df_test.to_csv(path_to_save / "test.csv", index=False)
 
 
-def load_cola_data_file_input(
+def load_sst2_data_file_input(
     path_to_train: Path, path_to_val: Path, path_to_test: Path
 ):
     path_to_train.parent.mkdir(parents=True, exist_ok=True)
     path_to_val.parent.mkdir(parents=True, exist_ok=True)
     path_to_test.parent.mkdir(parents=True, exist_ok=True)
 
-    df_train, df_val, df_test = _get_cola_data()
+    df_train, df_val, df_test = _get_sst2_data()
 
     df_train.to_csv(path_to_train, index=False)
     df_val.to_csv(path_to_val, index=False)

module-3/generative-example/Makefile

@@ -8,25 +8,12 @@ run_dev_gpu: build
 	docker run --net=host --gpus all -it -v ${PWD}:/main generative-example:latest /bin/bash
 
 format:
-	ruff format nlp_sample/ tests/
+	ruff format nlp_sample/ 
 
 lint:
-	ruff check nlp_sample/ tests/
+	ruff check nlp_sample/
 
-test:
-	pytest --disable-warnings ./tests/
 
-test_data:
-	pytest --disable-warnings -rc ./tests/test_data.py
-
-test_code:
-	pytest --disable-warnings -rc ./tests/test_code.py
-
-test_model:
-	pytest --disable-warnings -ss ./tests/test_model.py
-
-test_all:
-	pytest --cov=generative_example tests/
 
 train_example:
 	python generative_example/cli.py load-sql-data --subsample 0.1 ./data 

module-3/generative-example/README.md

@@ -22,7 +22,7 @@ export WANDB_API_KEY=***********************
 make test
 ```
 
-reference: https://madewithml.com/courses/mlops/testing/
+reference: https://docs.confident-ai.com/docs/confident-ai-github-actions
 
 ## Reports
 

module-3/generative-example/generative_example/config.py

@@ -1,5 +1,4 @@
 from dataclasses import dataclass
-from typing import Optional
 
 
 @dataclass

module-3/generative-example/generative_example/predictor.py

@@ -22,7 +22,7 @@ def __init__(self, model_load_path: str):
             model_load_path,
             low_cpu_mem_usage=True,
             return_dict=True,
-            torch_dtype=torch.bfloat16, #torch.float16,
+            torch_dtype=torch.bfloat16,
             trust_remote_code=True,
             device_map=device_map,
         )

module-3/generative-example/tests/conftest.py

@@ -1,32 +0,0 @@
-from pathlib import Path
-from typing import Tuple
-
-import great_expectations as ge
-import pandas as pd
-import pytest
-from great_expectations.dataset.pandas_dataset import PandasDataset
-
-from nlp_sample.data import load_cola_data
-
-
-@pytest.fixture(scope="session")
-def data_path() -> Path:
-    _data_path = Path("/tmp/data")
-    _data_path.mkdir(exist_ok=True, parents=True)
-
-    load_cola_data(path_to_save=_data_path)
-
-    return _data_path
-
-
-@pytest.fixture(scope="session")
-def data(data_path: Path) -> Tuple[PandasDataset, PandasDataset]:
-    df_train = pd.read_csv(data_path / "train.csv")
-    df_val = pd.read_csv(data_path / "val.csv")
-    df_test = pd.read_csv(data_path / "test.csv")
-
-    return (
-        ge.dataset.PandasDataset(df_train),
-        ge.dataset.PandasDataset(df_val),
-        ge.dataset.PandasDataset(df_test),
-    )