@codin-io/mlxd 1.0.17

package/mlxd/index.bundle.js.LICENSE.txt

@@ -0,0 +1,6 @@
+/* @license
+Papa Parse
+v5.4.1
+https://github.com/mholt/PapaParse
+License: MIT
+*/

package/mlxd/python/model_inference.py

@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+import sys
+import json
+import numpy as np
+from sklearn.metrics import accuracy_score, f1_score, mean_squared_error
+
+# Optional: import real ML libraries
+# from transformers import AutoModelForSequenceClassification, AutoTokenizer
+# import torch
+
+
+def load_dataset(data_path: str):
+    # Replace this logic to load an actual dataset (CSV, JSON, etc.)
+    # Here, we're just simulating random predictions & labels for demo purposes.
+    preds = np.random.randint(0, 2, 100)
+    labels = np.random.randint(0, 2, 100)
+    return preds, labels
+
+
+def validate_dataset(preds, labels):
+    # Example: Check if we got enough samples or if data is empty.
+    if len(preds) == 0 or len(labels) == 0:
+        return False, "No data found for predictions or labels."
+    if len(preds) != len(labels):
+        return False, "Predictions and labels have different lengths."
+    return True, ""
+
+
+def load_model(model_path: str):
+    # In a real scenario, load your ML model here (e.g., GPT-Neo, BERT).
+    # Example (commented out):
+    # model = AutoModelForSequenceClassification.from_pretrained(model_path)
+    # tokenizer = AutoTokenizer.from_pretrained(model_path)
+    return None
+
+
+def main():
+    # CLI arguments:
+    # 1) Metric name (e.g., "accuracy", "f1", "rmse").
+    # 2) Optional dataset path (default: "my_real_dataset.csv").
+    # 3) Optional model path (default: None).
+    metric_name = sys.argv[1] if len(sys.argv) > 1 else "accuracy"
+    data_path = sys.argv[2] if len(sys.argv) > 2 else "my_real_dataset.csv"
+    model_path = sys.argv[3] if len(sys.argv) > 3 else None
+
+    # Load data (predictions & labels) and validate.
+    preds, labels = load_dataset(data_path)
+    passed, message = validate_dataset(preds, labels)
+    if not passed:
+        print(json.dumps({"error": f"Data validation failed: {message}"}))
+        sys.exit(1)
+
+    # Load the model if needed (inference logic not shown in this sample).
+    model = load_model(model_path)
+
+    # Calculate the requested metric.
+    if metric_name == "accuracy":
+        value = accuracy_score(labels, preds)
+    elif metric_name == "f1":
+        value = f1_score(labels, preds)
+    elif metric_name == "rmse":
+        value = mean_squared_error(labels, preds, squared=False)
+    else:
+        value = 0.0
+
+    # Output metric as JSON for the CLI to parse.
+    print(json.dumps({"metric": metric_name, "value": float(value)}))
+
+
+if __name__ == "__main__":
+    main()

package/mlxd/python/python/model_inference.py

@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+import sys
+import json
+import numpy as np
+from sklearn.metrics import accuracy_score, f1_score, mean_squared_error
+
+# Optional: import real ML libraries
+# from transformers import AutoModelForSequenceClassification, AutoTokenizer
+# import torch
+
+
+def load_dataset(data_path: str):
+    # Replace this logic to load an actual dataset (CSV, JSON, etc.)
+    # Here, we're just simulating random predictions & labels for demo purposes.
+    preds = np.random.randint(0, 2, 100)
+    labels = np.random.randint(0, 2, 100)
+    return preds, labels
+
+
+def validate_dataset(preds, labels):
+    # Example: Check if we got enough samples or if data is empty.
+    if len(preds) == 0 or len(labels) == 0:
+        return False, "No data found for predictions or labels."
+    if len(preds) != len(labels):
+        return False, "Predictions and labels have different lengths."
+    return True, ""
+
+
+def load_model(model_path: str):
+    # In a real scenario, load your ML model here (e.g., GPT-Neo, BERT).
+    # Example (commented out):
+    # model = AutoModelForSequenceClassification.from_pretrained(model_path)
+    # tokenizer = AutoTokenizer.from_pretrained(model_path)
+    return None
+
+
+def main():
+    # CLI arguments:
+    # 1) Metric name (e.g., "accuracy", "f1", "rmse").
+    # 2) Optional dataset path (default: "my_real_dataset.csv").
+    # 3) Optional model path (default: None).
+    metric_name = sys.argv[1] if len(sys.argv) > 1 else "accuracy"
+    data_path = sys.argv[2] if len(sys.argv) > 2 else "my_real_dataset.csv"
+    model_path = sys.argv[3] if len(sys.argv) > 3 else None
+
+    # Load data (predictions & labels) and validate.
+    preds, labels = load_dataset(data_path)
+    passed, message = validate_dataset(preds, labels)
+    if not passed:
+        print(json.dumps({"error": f"Data validation failed: {message}"}))
+        sys.exit(1)
+
+    # Load the model if needed (inference logic not shown in this sample).
+    model = load_model(model_path)
+
+    # Calculate the requested metric.
+    if metric_name == "accuracy":
+        value = accuracy_score(labels, preds)
+    elif metric_name == "f1":
+        value = f1_score(labels, preds)
+    elif metric_name == "rmse":
+        value = mean_squared_error(labels, preds, squared=False)
+    else:
+        value = 0.0
+
+    # Output metric as JSON for the CLI to parse.
+    print(json.dumps({"metric": metric_name, "value": float(value)}))
+
+
+if __name__ == "__main__":
+    main()

package/package.json

@@ -0,0 +1,56 @@
+{
+  "name": "@codin-io/mlxd",
+  "version": "1.0.17",
+  "description": "CI/CD Testing Tool for ML",
+  "author": "codin.io",
+  "license": "ISC",
+  "main": "mlxd/index.js",
+  "bin": {
+    "mlxd": "mlxd/index.bundle.js"
+  },
+  "scripts": {
+    "build": "webpack && cp -r python mlxd/python && cp config.yaml mlxd/ && cp README.md mlxd/",
+    "test": "jest",
+    "postbuild": "node add-shebang.js",
+    "prepublishOnly": "npm run build",
+    "dev": "ts-node src/index.ts",
+    "postinstall": "node scripts/moveMlxd.js"
+  },
+  "files": [
+    "mlxd/**/*",
+    "python/**/*",
+    "config.yaml",
+    "README.md",
+    "LICENSE"
+  ],
+  "devDependencies": {
+    "@types/axios": "^0.14.4",
+    "@types/dotenv": "^8.2.3",
+    "@types/jest": "^29.5.14",
+    "@types/js-yaml": "^4.0.9",
+    "@types/markdown-pdf": "^9.0.5",
+    "@types/mocha": "^10.0.10",
+    "@types/node": "^22.10.2",
+    "@types/nodemailer": "^6.4.17",
+    "@types/papaparse": "^5.3.15",
+    "jest": "^29.7.0",
+    "terser": "^5.37.0",
+    "terser-webpack-plugin": "^5.3.11",
+    "ts-jest": "^29.2.5",
+    "ts-loader": "^9.5.1",
+    "ts-node": "^10.9.2",
+    "typescript": "^4.7.4",
+    "webpack": "^5.97.1",
+    "webpack-cli": "^6.0.1",
+    "webpack-obfuscator": "^3.5.1"
+  },
+  "dependencies": {
+    "axios": "^1.7.9",
+    "dotenv": "^16.4.7",
+    "handlebars": "^4.7.8",
+    "js-yaml": "^4.1.0",
+    "markdown-pdf": "^11.0.0",
+    "nodemailer": "^6.9.16",
+    "papaparse": "^5.4.1"
+  }
+}

package/python/model_inference.py

@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+import sys
+import json
+import numpy as np
+from sklearn.metrics import accuracy_score, f1_score, mean_squared_error
+
+# Optional: import real ML libraries
+# from transformers import AutoModelForSequenceClassification, AutoTokenizer
+# import torch
+
+
+def load_dataset(data_path: str):
+    # Replace this logic to load an actual dataset (CSV, JSON, etc.)
+    # Here, we're just simulating random predictions & labels for demo purposes.
+    preds = np.random.randint(0, 2, 100)
+    labels = np.random.randint(0, 2, 100)
+    return preds, labels
+
+
+def validate_dataset(preds, labels):
+    # Example: Check if we got enough samples or if data is empty.
+    if len(preds) == 0 or len(labels) == 0:
+        return False, "No data found for predictions or labels."
+    if len(preds) != len(labels):
+        return False, "Predictions and labels have different lengths."
+    return True, ""
+
+
+def load_model(model_path: str):
+    # In a real scenario, load your ML model here (e.g., GPT-Neo, BERT).
+    # Example (commented out):
+    # model = AutoModelForSequenceClassification.from_pretrained(model_path)
+    # tokenizer = AutoTokenizer.from_pretrained(model_path)
+    return None
+
+
+def main():
+    # CLI arguments:
+    # 1) Metric name (e.g., "accuracy", "f1", "rmse").
+    # 2) Optional dataset path (default: "my_real_dataset.csv").
+    # 3) Optional model path (default: None).
+    metric_name = sys.argv[1] if len(sys.argv) > 1 else "accuracy"
+    data_path = sys.argv[2] if len(sys.argv) > 2 else "my_real_dataset.csv"
+    model_path = sys.argv[3] if len(sys.argv) > 3 else None
+
+    # Load data (predictions & labels) and validate.
+    preds, labels = load_dataset(data_path)
+    passed, message = validate_dataset(preds, labels)
+    if not passed:
+        print(json.dumps({"error": f"Data validation failed: {message}"}))
+        sys.exit(1)
+
+    # Load the model if needed (inference logic not shown in this sample).
+    model = load_model(model_path)
+
+    # Calculate the requested metric.
+    if metric_name == "accuracy":
+        value = accuracy_score(labels, preds)
+    elif metric_name == "f1":
+        value = f1_score(labels, preds)
+    elif metric_name == "rmse":
+        value = mean_squared_error(labels, preds, squared=False)
+    else:
+        value = 0.0
+
+    # Output metric as JSON for the CLI to parse.
+    print(json.dumps({"metric": metric_name, "value": float(value)}))
+
+
+if __name__ == "__main__":
+    main()