Free Notebook Usage
Free Notebooks are like "free trial versions," allowing you to experience the features of Notebooks and see if they suit your needs. Although there are some limitations, they are more than sufficient for learning and simple projects.
What is included in the free version?
Basic Resources
Compute Resources:
| Resource Type | Specification | Image Resources |
|---|---|---|
| CPU | CPU basic · 0.5v CPU · 500MB | ubuntu22-python3.10-jupyter-cpu:v1.0.1-notebook ubuntu22-vllm-python3.10-jupyter-cpu:v1.0.1-notebook ubuntu22-sglang-python3.10-jupyter-cpu:v1.0.1-notebook |
| CPU | CPU basic · 2v CPU · 4GB | Same as above |
| CPU | CPU basic · 4v CPU · 8GB | Same as above |
| CPU | CPU basic · 8v CPU · 16GB | Same as above |
| CPU | CPU basic · 16v CPU · 32GB | Same as above |
| NPU | NPU basic · 1 × NPU 910B · 4v CPU · 8GB | ubuntu22-cann8.3-python3.10-jupyter:v4.0.1-notebook ubuntu22-cann8.3-sglang-python3.10-jupyter:v4.0.1-notebook ubuntu22-cann8.3-vllm-python3.10-jupyter:v4.0.1-notebook ubuntu22-cann8.5-python3.11-jupyter:v1.0.1-notebook ubuntu22-cann8.5-python3.11-vllm-jupyter:v1.0.0-notebook ubuntu22-cann8.5-python3.11-sglang-jupyter:v1.0.1-notebook |
| NPU | NPU basic · 1 × NPU 910B · 16v CPU · 32GB | Same as above |
Software Environment: Python 3.11 version, pre-installed with common data science packages, allows installation of required packages, supports multiple programming languages.
What are the usage restrictions?
Resource Limitations
Running Limitations: Only one Notebook can be running at a time, with a maximum runtime of 2 hours each time, and a monthly limit of 1000 core hours. The core hour consumption depends on the selected specification (for example, selecting the CPU basic · 4v CPU · 8GB specification, using it for 1 hour would consume 4 core hours).
Storage Limitations: Workspace storage is limited to 50GB, temporary storage is limited to 2GB, individual files cannot exceed 100MB, and a maximum of 10 Notebooks can be created.
Functional Limitations
Usage: A running Notebook cannot be paused. If it times out or encounters an error, it will automatically terminate. If not started successfully, no core hours will be consumed. To close a running Notebook, you must do so from the start page (Workbench -> My Notebooks).
API Usage: Maximum of 60 requests per minute, daily data transfer limit of 1GB, and a maximum of 5 simultaneous requests.
Collaboration Features: Up to 3 people can view, only read-only versions can be shared, comments can be added, and only the last 5 versions are retained.
Getting Started
Registration Process
Create an Account: Visit the registration page, fill in basic information, verify your email, and complete the setup.
Environment Initialization: Use the command line tool to initialize the workspace and verify the Python environment.
Basic Operations
Create a Notebook: Create a new Notebook using the command line tool, and import example projects.
Manage Files: Supports uploading and downloading files, managing project data.
Resource Optimization
Memory Management
-
Optimization Tips
import gc
import torch
def optimize_memory():
# Clean up unused objects
gc.collect()
# Clear PyTorch cache
torch.cuda.empty_cache()
# Use generators for large data
def data_generator():
for chunk in pd.read_csv("large_file.csv", chunksize=1000):
yield process_chunk(chunk) -
Data Processing
# Process large files in chunks
def process_large_file(file_path):
results = []
for chunk in pd.read_csv(file_path, chunksize=1000):
result = process_chunk(chunk)
results.append(result)
return pd.concat(results)
Compute Optimization
-
Parallel Processing
from multiprocessing import Pool
def parallel_process(data_list):
with Pool(processes=2) as pool:
results = pool.map(process_function, data_list)
return results -
Code Optimization
# Use vectorized operations
import numpy as np
def optimize_calculation(data):
# Replace loop operations
result = np.vectorize(process_function)(data)
return result
Usage Tips
Improve Efficiency
Code Organization
- Divide code into different functions
- Each function should perform one task
- Keep code structure clear
- Easy to understand and maintain
Use Cache
- Use cache for repeated calculations
- Avoid running the same code repeatedly
- Save time and resources
- Improve execution efficiency
Storage Management
File Management
- Regularly clean up temporary files
- Compress old data files
- Delete unnecessary files
- Keep the workspace tidy
Data Optimization: Choose appropriate data types, compress large datasets, optimize storage formats to save storage space.
When to Upgrade?
Upgrade Signals
Resource Shortage: CPU usage frequently exceeds 80%, memory usage frequently exceeds 75%, storage usage exceeds 90%, and often encounter various restrictions.
Functional Requirements: Need GPU acceleration, require more storage space, need longer runtime, need more collaboration features.
Upgrade Steps
Preparation: Evaluate actual needs, select an appropriate paid plan, prepare data migration, configure the new environment.
Data Migration: Export the current workspace, migrate to the new environment, verify data integrity, and start using new features.
Frequently Asked Questions
Resource Issues
How to check resource usage? Check the resource monitoring in the Notebook interface, use system commands to check CPU and memory, regularly check storage space usage.
How to handle insufficient memory? Close unnecessary Notebooks, clean up temporary variables and data, use smaller datasets, consider upgrading to a paid version. How to optimize runtime? Use efficient algorithms, avoid redundant calculations, use caching reasonably, optimize data processing workflows.
How to manage storage space? Regularly clean up unnecessary files, compress large datasets, use external storage services, delete temporary files promptly.
Function Issues
How to install additional packages? Install Python packages using pip, install scientific computing packages using conda, check package compatibility, note storage space limitations.
How to share a Notebook? Set sharing permissions, generate a sharing link, invite collaborators, control access permissions.
Limitation Issues
Runtime Limitations: Maximum runtime is 1 hour, auto-stops after 30 minutes of inactivity, arrange computing tasks properly, save work results in time.
Storage Space Limitations: Workspace storage is limited to 10GB, individual files cannot exceed 100MB, regularly clean up and compress, consider upgrading to a paid version.
Summary
Free Notebooks are a great choice to start learning AI. By using them wisely, you can experience for free (use Notebooks without paying), learn the basics (master basic programming and data processing skills), practise projects (complete small AI projects), and assess needs (understand whether you need to upgrade to a paid version).
Remember, although the free version has limitations, it is more than enough for learning and simple projects. Plan properly and optimize your usage to make full use of free resources!