Generate AI-Powered Love Poems in One Click with Huawei ModelArts

Tomorrow is Qixi, and this guide by B‑station creator Xiao Pei walks developers through using Huawei Cloud ModelArts to build an AI that generates love poems.

Step‑by‑step implementation

№1. Log in to the Huawei Cloud website and navigate to ModelArts under EI Enterprise Intelligence.

№2. Open the console.

№3. Click the "User Guide" in the upper‑right corner.

№4. Open "My Credentials".

№5. Add a new Access Key.

№6. Return to the console and select "Global Configuration".

№7. Add the Access Key you just created.

№8. In the development environment, open Notebook.

№9. Create a Notebook environment.

№10. After the environment starts, open the Notebook.

№11. Open a terminal and run conda init to initialize the environment.

№12. Run conda info --envs to view available Conda environments.

№13. Activate the TensorFlow environment.

№14. Install Keras.

№15. Install TensorFlow 2.2.0.

№16. Create a new Notebook script and select the TensorFlow environment.

№17. Import the required packages and modules.

№18. Write text‑processing functions.

№19. Upload the dataset files.

№20. Choose the Shakespeare poetry dataset.

№21. Create a folder named models.

№22. Upload the pre‑trained model files into the models folder.

№23. Set parameters and load the pre‑trained model.

№24. Train the model (if needed).

№25. After training, provide a seed text and start generating poetry.

№26. The AI finishes generating the poem.

This tutorial equips developers with a complete workflow—from environment setup to model training and poem generation—enabling them to create AI‑driven love poems for the Qixi festival.