added some tests for minGPT pipeline

.github/workflows/python-tests.yml

@@ -0,0 +1,32 @@
+name: Run Python Tests
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Check out the code
+        uses: actions/checkout@v2
+
+      - name: Set up Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: '3.8'  # Adjust the Python version if needed
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -r requirements.txt  # Ensure you have a `requirements.txt` file for dependencies
+
+      - name: Run tests
+        run: |
+          python -m unittest discover -s minGPT/tests -p "*.py"
+

minGPT/tests/test_data/test_data.csv

@@ -0,0 +1,16 @@
+mol_smiles	conductivity
+NC(=O)CSCC(CO[Cu])OC(=O)[Au]	1
+CCC(F)C(=O)NC(CO[Cu])COC(=O)[Au]	0
+CCSCCN(CCN[Cu])CCOC(=O)[Au]	1
+C#CCN(CCOCCO[Cu])CCOC(=O)[Au]	1
+CCC(COC(=O)[Au])C(=O)NCCO[Cu]	0
+CC(CCOC(=O)[Au])SCCO[Cu]	1
+CN(CCNC(=O)COC(=O)[Au])CCO[Cu]	1
+CN(CCO[Cu])C(=O)CCOC(=O)[Au]	1
+CCC(C)C(NC(=O)[Au])C(=O)NC(C)CN[Cu]	0
+CCS(C)(=O)=NC(=O)C(CN[Cu])OC(=O)[Au]	0
+C=CC(CNCCCOCCOC(=O)[Au])O[Cu]	1
+CCC(=O)CNC(=O)C(CN[Cu])OC(=O)[Au]	0
+CC(CCO[Cu])N(C)CCCCCOC(=O)[Au]	1
+CC(C)CNC(CO[Cu])COC(=O)[Au]	0
+NC(=O)CCC(=O)NCC(CO[Cu])OC(=O)[Au]	1

minGPT/tests/test_pipeline.py

@@ -0,0 +1,89 @@
+import unittest
+import sys
+import os
+import pandas as pd
+from minGPT.pipeline import minGPT
+
+class TestMinGPTPipeline(unittest.TestCase):
+
+    def setUp(self):
+        self.pipeline = minGPT()
+        self.data_file_path = os.path.join("minGPT", "tests", "test_data", "test_data.csv")
+        self.model_config = self.pipeline.get_default_model_config()
+        self.data_config = self.pipeline.get_default_data_config()
+        self.data_config.file_path = self.data_file_path
+
+    def load_model_with_config(self, train_dataset):
+        """Helper method to load model with vocab and block sizes."""
+        self.model_config.vocab_size = train_dataset.get_vocab_size()
+        self.model_config.block_size = train_dataset.get_block_size()
+        self.pipeline.load_model(self.model_config)
+
+    def test_data_preprocessing(self):
+        data_config = self.pipeline.get_default_data_config()
+        data_config.file_path = self.data_file_path
+        data_config.block_size = 64
+
+        raw_data = pd.read_csv(self.data_file_path, sep="\t")
+        print("Unprocessed Data (First 5 Rows):")
+        print(raw_data.head())
+
+        train_dataset, test_dataset = self.pipeline.data_preprocessing(data_config)
+
+        self.assertIsNotNone(train_dataset, "Train dataset should not be None")
+        self.assertIsNotNone(test_dataset, "Test dataset should not be None")
+        self.assertGreater(len(train_dataset), 0, "Train dataset should contain data")
+        self.assertGreater(len(test_dataset), 0, "Test dataset should contain data")
+
+    def test_model_loading(self):
+        train_dataset, _ = self.pipeline.data_preprocessing(self.data_config)
+        self.load_model_with_config(train_dataset)
+        self.assertIsNotNone(self.pipeline.model, "Model should be loaded")
+
+    def test_training(self):
+        train_config = self.pipeline.get_default_train_config()
+        train_config.max_iters = 10
+        train_config.device = 'cpu'
+
+        train_dataset, _ = self.pipeline.data_preprocessing(self.data_config)
+        self.load_model_with_config(train_dataset)
+
+        def batch_end_callback(trainer):
+            print(f"Batch {trainer.iter_num} ended with loss: {trainer.loss.item() if hasattr(trainer, 'loss') else 'N/A'}")
+
+        train_config.call_back = batch_end_callback
+        loss = self.pipeline.train(train_config)
+        self.assertIsNotNone(loss, "Training loss should not be None")
+
+    def test_generation(self):
+        generate_config = self.pipeline.get_default_generate_config()
+        generate_config.num_samples = 5
+        generate_config.ckpt_path = os.path.join("minGPT", "ckpts", "sample_checkpoint.pt")
+
+        train_dataset, _ = self.pipeline.data_preprocessing(self.data_config)
+        self.load_model_with_config(train_dataset)
+
+        generated_out = self.pipeline.generate(generate_config)
+        self.assertIsInstance(generated_out, list, "Generation output should be a list")
+        self.assertGreater(len(generated_out), 0, "Generated output should contain elements")
+        print("Generated Output:", generated_out)
+
+    def test_evaluation(self):
+        train_dataset, _ = self.pipeline.data_preprocessing(self.data_config)
+        self.pipeline.df = pd.DataFrame({"mol_smiles": ["C(=O)CSCC(CO[Cu])OC(=O)[Au]"]})
+
+        generate_config = self.pipeline.get_default_generate_config()
+        generate_config.num_samples = 5
+        generate_config.ckpt_path = os.path.join("minGPT", "ckpts", "sample_checkpoint.pt")
+
+        self.load_model_with_config(train_dataset)
+        generated_out = self.pipeline.generate(generate_config)
+        self.pipeline.gen_df = pd.DataFrame({"mol_smiles": generated_out})
+
+        results = self.pipeline.evaluate()
+        self.assertIsInstance(results, tuple, "Evaluation should return a tuple")
+        self.assertEqual(len(results), 6, "Evaluation should return six metrics")
+        print("Evaluation Results:", results)
+
+if __name__ == '__main__':
+    unittest.main()

minGPT_pipeline.ipynb

@@ -2,10 +2,27 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 46,
+   "execution_count": 1,
    "id": "5593892c",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "No normalization for SPS. Feature removed!\n",
+      "No normalization for AvgIpc. Feature removed!\n",
+      "/Users/arashkajeh/Documents/GitHub/PolyGen/venv38/lib/python3.8/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n",
+      "Skipped loading some Tensorflow models, missing a dependency. No module named 'tensorflow'\n",
+      "Skipped loading modules with pytorch-geometric dependency, missing a dependency. No module named 'torch_geometric'\n",
+      "Skipped loading modules with pytorch-geometric dependency, missing a dependency. cannot import name 'DMPNN' from 'deepchem.models.torch_models' (/Users/arashkajeh/Documents/GitHub/PolyGen/venv38/lib/python3.8/site-packages/deepchem/models/torch_models/__init__.py)\n",
+      "Skipped loading modules with pytorch-lightning dependency, missing a dependency. No module named 'lightning'\n",
+      "Skipped loading some Jax models, missing a dependency. No module named 'jax'\n",
+      "Skipped loading some PyTorch models, missing a dependency. No module named 'tensorflow'\n"
+     ]
+    }
+   ],
    "source": [
     "import pandas as pd\n",
     "import numpy as np\n",
@@ -29,7 +46,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 47,
+   "execution_count": 2,
    "id": "0b36bb88-5c31-47ab-bbfc-9f283898c5e8",
    "metadata": {},
    "outputs": [
@@ -59,7 +76,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 48,
+   "execution_count": 3,
    "id": "66681958-a1f1-4fcd-8216-e2a15fc54c7c",
    "metadata": {},
    "outputs": [],
@@ -69,7 +86,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 49,
+   "execution_count": 4,
    "id": "9bdf2a24-8cd4-41c7-bedd-07525412f532",
    "metadata": {},
    "outputs": [
@@ -100,7 +117,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": null,
    "id": "29b97415-56ac-4b8d-bda6-214d93ab8e15",
    "metadata": {},
    "outputs": [
@@ -121,7 +138,15 @@
       "call_back: None\n",
       "pretrain: None\n",
       "\n",
-      "auto\n"
+      "auto\n",
+      "running on device cpu\n",
+      "iter_dt 0.00ms; iter 0: train loss 6.40999, val loss 6.41125\n",
+      "iter_dt 1052.81ms; iter 100: train loss 2.05009, val loss 2.04999\n",
+      "iter_dt 1192.84ms; iter 200: train loss 0.56169, val loss 0.56216\n",
+      "iter_dt 446.42ms; iter 300: train loss 0.34879, val loss 0.35418\n",
+      "iter_dt 1401.88ms; iter 400: train loss 0.30944, val loss 0.30891\n",
+      "iter_dt 786.11ms; iter 500: train loss 0.28507, val loss 0.28915\n",
+      "iter_dt 445.25ms; iter 600: train loss 0.26450, val loss 0.26419\n"
      ]
     }
    ],
@@ -145,7 +170,7 @@
     "\n",
     "train_config.call_back = batch_end_callback\n",
     "# Uncomment the following line to start training\n",
-    "# loss = pipeline.train(train_config)"
+    "loss = pipeline.train(train_config)"
    ]
   },
   {
@@ -158,23 +183,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 51,
+   "execution_count": null,
    "id": "f72b13fe-9653-4f6b-9d9b-cc4a00f51eea",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "ckpts_path: None\n",
-      "num_samples: 100\n",
-      "temperature: 1.0\n",
-      "task: conditional\n",
-      "ckpt_path: ./minGPT/ckpts/10000.pt\n",
-      "\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "generate_config = pipeline.get_default_generate_config()\n",
     "generate_config.ckpt_path = \"./minGPT/ckpts/10000.pt\"\n",
@@ -196,18 +208,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 52,
+   "execution_count": null,
    "id": "5c824f5e",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "(0.97, 0.5800000000000001, 0.87, 0.9772727272727273, 0.2633707205965721, 0.6905273247947646)\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "print(pipeline.evaluate())"
    ]
@@ -219,6 +223,14 @@
    "metadata": {},
    "outputs": [],
    "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7540c31b-4701-490c-856c-bdf85cffc99b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {
@@ -242,4 +254,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 5
-}
+}

requirements.txt

@@ -4,3 +4,4 @@ transformers==4.46.0
 ipykernel==6.29.5
 torch==2.0.1
 notebook==7.2.2
+pandas==2.0.3