Tutorial: Running inference with Zhipu AI's GLM-4.5 using W&B Inference

pip install openai wandb weave

# Set environment variables in your notebook
%env WANDB_API_KEY=your-wandb-api-key-here

export WANDB_API_KEY="your-wandb-api-key-here"

import os
# Set environment variables programmatically
os.environ["WANDB_API_KEY] = "your-wandb-api-key-here"

import os
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

response = client.chat.completions.create(
    model="zai-org/GLM-4.5",
    messages=[
        {"role": "system", "content": "You serve as an expert educator specializing in quantum computing, breaking down complex concepts into clear, accessible explanations that inspire curiosity and understanding."},
        {"role": "user", "content": "Explain the fundamental principles of quantum computing in an engaging way. Cover key concepts like superposition, entanglement, quantum gates, and qubits. Discuss how these elements work together to perform computations that classical computers cannot efficiently handle. Include practical examples and potential real-world applications that demonstrate quantum computing's transformative potential."}
    ],
    temperature=0.7,
)

print("Response:", response.choices[0].message.content)


import os
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

@weave.op()
def creative_writing_experiment():
    # Test different temperature settings
    temperatures = [0.1, 0.7, 1.2]

    for temp in temperatures:
        print(f"\n--- Temperature: {temp} ---")
        response = client.chat.completions.create(
            model="zai-org/GLM-4.5",
            messages=[
                {"role": "system", "content": "You are a creative writing coach who helps writers develop compelling narratives, explore different styles, and craft engaging stories that resonate with readers."},
                {"role": "user", "content": "Write a short story about a mysterious antique shop where time seems to stand still. The owner has a secret, and customers often leave with more than just purchases. Focus on atmosphere, subtle mystery, and the interplay between past and present. Keep the story under 500 words."}
            ],
            temperature=temp,
            max_tokens=400,
        )
        print(response.choices[0].message.content)

result = creative_writing_experiment()

import os
import sys
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

stream = client.chat.completions.create(
    model="zai-org/GLM-4.5",
    messages=[
        {"role": "system", "content": "You are a compelling narrative architect who crafts engaging stories about humanity's greatest achievements, blending historical facts with inspiring human elements."},
        {"role": "user", "content": "Craft an inspiring narrative about humanity's journey into space, from ancient stargazers to modern space explorers. Highlight pivotal moments, courageous individuals, technological innovations, international collaborations, and the philosophical implications of becoming a multi-planetary species. Weave together scientific achievements with the human spirit of discovery and the challenges overcome along the way."}
    ],
    stream=True,
    temperature=0.7,
)

sys.stdout.write("Response: ")
for chunk in stream:
    delta = chunk.choices[0].delta
    if delta and delta.content:
        sys.stdout.write(delta.content)
        sys.stdout.flush()
print()


import os
import openai
import weave
from pypdf import PdfReader

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

def extract_pdf_text(file_path):
    reader = PdfReader(file_path)
    text = ""
    for page in reader.pages:
        text += page.extract_text() + "\n"
    return text

@weave.op()
def analyze_technical_documentation():
    # For demo purposes, we'll use a sample text instead of reading a PDF
    sample_text = """
    AWS Cloud Computing Services Overview:

    Amazon Web Services (AWS) provides a comprehensive suite of cloud computing services designed to help businesses scale and innovate. Key services include:

    1. Compute Services:
    - Amazon EC2: Virtual servers in the cloud
    - AWS Lambda: Serverless compute service
    - Amazon ECS/EKS: Container orchestration

    2. Storage Services:
    - Amazon S3: Object storage service
    - Amazon EBS: Block storage for EC2 instances
    - Amazon Glacier: Long-term archival storage

    3. Database Services:
    - Amazon RDS: Managed relational databases
    - Amazon DynamoDB: NoSQL database service
    - Amazon Redshift: Data warehousing solution

    These services enable organizations to build resilient, scalable, and cost-effective cloud architectures.
    """

    response = client.chat.completions.create(
        model="zai-org/GLM-4.5",
        messages=[
            {"role": "system", "content": "You are a technical documentation specialist who excels at analyzing complex technical materials, extracting key insights, and providing clear summaries with actionable recommendations."},
            {"role": "user", "content": f"Analyze this technical documentation and provide a detailed summary covering the main services, their purposes, and how they work together to create comprehensive cloud solutions. Identify key benefits, potential use cases, and architectural considerations. Documentation content: {sample_text}"}
        ],
        temperature=0.4,
    )
    return response.choices[0].message.content

result = analyze_technical_documentation()
print("Technical Documentation Analysis:", result)

import os
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

@weave.op()
def translate_and_explain():
    response = client.chat.completions.create(
        model="zai-org/GLM-4.5",
        messages=[
            {"role": "system", "content": "You are a proficient multilingual communicator skilled in translation, cultural adaptation, and cross-language communication strategies."},
            {"role": "user", "content": "Translate the following English text to Spanish, French, and German. For each translation, provide cultural context and explain any idiomatic expressions or cultural nuances that might affect the translation. Also suggest appropriate communication strategies for each language when discussing this topic. Text to translate: 'The key to successful international business is understanding cultural differences and adapting communication styles accordingly. Building trust through respectful dialogue and genuine interest in local customs can open doors to meaningful partnerships worldwide.'"}
        ],
        temperature=0.5,
    )
    return response.choices[0].message.content

result = translate_and_explain()
print("Multilingual Translation and Analysis:", result)


import os
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

@weave.op()
def analyze_saas_business_model():
    response = client.chat.completions.create(
        model="zai-org/GLM-4.5",
        messages=[
            {"role": "system", "content": "You are a seasoned SaaS strategist who provides deep insights into software business models, market positioning, and growth strategies for subscription-based companies."},
            {"role": "user", "content": "Analyze the SaaS business model for a project management platform. Provide a comprehensive assessment covering: 1) Key value propositions and competitive advantages, 2) Pricing strategy recommendations with different tiers, 3) Customer acquisition and retention strategies, 4) Key metrics to track (MRR, churn rate, CAC, LTV), 5) Scaling challenges and solutions, 6) Market positioning against competitors like Asana, Trello, and Monday.com. Include specific recommendations for product development and go-to-market strategies."}
        ],
        temperature=0.6,
    )
    return response.choices[0].message.content

result = analyze_saas_business_model()
print("SaaS Business Analysis:", result)

import os
import openai
import weave

PROJECT = "wandb_inference"
weave.init(PROJECT)

client = openai.OpenAI(
    base_url="https://api.inference.wandb.ai/v1",
    api_key=os.getenv("WANDB_API_KEY"),
    project=PROJECT,
    default_headers={
        "OpenAI-Project": "wandb_fc/quickstart_playground"  # replace with your actual team/project
    }
)

@weave.op()
def analyze_customer_feedback():
    response = client.chat.completions.create(
        model="zai-org/GLM-4.5",
        messages=[
            {"role": "system", "content": "You are a skilled sentiment analyst who examines customer feedback with precision, identifying emotional tones, key concerns, and actionable insights to improve user experience."},
            {"role": "user", "content": "Analyze the following customer reviews and provide a comprehensive sentiment breakdown. Identify positive aspects, areas for improvement, recurring themes, and specific recommendations for enhancing customer satisfaction. Reviews: 1. 'The product works great but customer service was slow to respond.' 2. 'Amazing quality and fast shipping, will definitely buy again!' 3. 'Interface is confusing and not user-friendly at all.' 4. 'Good value for money, exceeded my expectations.' 5. 'Delivery was delayed and packaging was damaged upon arrival.'"}
        ],
        temperature=0.3,
    )
    return response.choices[0].message.content

result = analyze_customer_feedback()
print("Customer Feedback Analysis:", result)