AI: 生成式 AI 与大模型

from openai import OpenAI
client = OpenAI(api_key="your-api-key")
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "How to return a product?"}]
)
print(response.choices[0].message.content)

# Minimal OpenAI API call
from openai import OpenAI
client = OpenAI()  # Uses OPENAI_API_KEY env variable
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Say hello in 3 languages."}]
)
print(response.choices[0].message.content)

Hello! / ¡Hola! / Hello!

Input:  "The cat sat on the"
Step 1: "The cat sat on the" → predict "mat"  (most likely next token)
Step 2: "The cat sat on the mat" → predict "."
Step 3: "The cat sat on the mat." → predict "<END>"
Output: "The cat sat on the mat."

Text:   "Hello, world!"
Tokens: ["Hello", ",", " world", "!"]     → 4 tokens

Text:   "artificial intelligence"
Tokens: ["art", "ific", "ial"]               → 3 tokens (subword tokenization)

# Calculate token count using tiktoken
import tiktoken

encoding = tiktoken.encoding_for_model("gpt-4o-mini")

texts = [
    "Hello, world!",
    "The quick brown fox jumps over the lazy dog.",
    "Artificial intelligence is changing the world.",
    "Generative AI is transforming the world."
]

for text in texts:
    tokens = encoding.encode(text)
    print(f"Text: {text}")
    print(f"  Tokens: {len(tokens)} | Decoded: {encoding.decode(tokens[:3])}...")
    print()

Text: Hello, world!
  Tokens: 3 | Decoded: Hello...

Text: The quick brown fox jumps over the lazy dog.
  Tokens: 10 | Decoded: The quick brown...

Text: Artificial intelligence is changing the world.
  Tokens: 8 | Decoded: Artifici...

Text: Generative AI is transforming the world.
  Tokens: 8 | Decoded: Generative AI is...

Temperature = 0    → Always pick the most likely token (deterministic)
Temperature = 0.7  → Mostly likely but with some variation (default for chat)
Temperature = 1.0  → Natural variation, follows the learned distribution
Temperature = 2.0  → Very random, often nonsensical

Top-P = 0.1  → Only consider top 10% most likely tokens (conservative)
Top-P = 0.9  → Consider tokens covering 90% probability (moderate)
Top-P = 1.0  → Consider all tokens (no filtering)

# Compare outputs at different temperatures
from openai import OpenAI
client = OpenAI()

prompt = "Write a one-sentence description of a sunset."
temperatures = [0, 0.5, 1.0, 1.5]

for temp in temperatures:
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}],
        temperature=temp,
        max_tokens=50
    )
    text = response.choices[0].message.content
    print(f"Temperature={temp}: {text}")
    print()

Temperature=0: The sun dipped below the horizon, painting the sky in shades of orange and pink.

Temperature=0.5: Golden light spilled across the horizon as the sun melted into a sea of amber clouds.

Temperature=1.0: The day's final breath set the clouds ablaze, draping the world in a warm, fading embrace of crimson and gold.

Temperature=1.5: Colors—copper, violet, impossible rose—bloomed like dying flowers across the fractured sky as the sun surrendered.

# Basic OpenAI API call with system prompt
from openai import OpenAI
client = OpenAI()

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[
        {"role": "system", "content": "You are a concise technical assistant."},
        {"role": "user", "content": "What is a Transformer in 2 sentences?"}
    ],
    temperature=0.3
)

print(response.choices[0].message.content)
print(f"\nTokens used: {response.usage.total_tokens}")

A Transformer is a neural network architecture based on self-attention, allowing it to process all input tokens in parallel rather than sequentially. Introduced in 2017, it became the foundation for modern LLMs like GPT and BERT.

Tokens used: 58

# Multi-turn conversation with OpenAI API
from openai import OpenAI
client = OpenAI()

conversation = [
    {"role": "system", "content": "You are a helpful Python tutor."}
]

questions = [
    "What is a list comprehension?",
    "Can you show me an example?",
    "What if I want to filter even numbers?"
]

for q in questions:
    conversation.append({"role": "user", "content": q})
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=conversation,
        temperature=0.3
    )
    reply = response.choices[0].message.content
    conversation.append({"role": "assistant", "content": reply})
    print(f"User: {q}")
    print(f"Bot:  {reply[:120]}...")
    print()

User: What is a list comprehension?
Bot:  A list comprehension is a concise way to create lists in Python using a single line of syntax, combining a for loop and optional conditions...

User: Can you show me an example?
Bot:  Sure! Here's a basic example: squares = [x**2 for x in range(10)] This creates a list [0, 1, 4, 9, 16, 25, 36, 49, 64, 81]....

User: What if I want to filter even numbers?
Bot:  Add a condition: evens = [x for x in range(20) if x % 2 == 0] This produces [0, 2, 4, 6, 8, 10, 12, 14, 16, 18]....

# Streaming output with OpenAI API
from openai import OpenAI
client = OpenAI()

stream = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Explain quantum computing in 3 sentences."}],
    temperature=0.5,
    stream=True
)

print("Streaming response: ", end="")
for chunk in stream:
    if chunk.choices[0].delta.content is not None:
        print(chunk.choices[0].delta.content, end="", flush=True)
print("\n\n[Stream complete]")

Streaming response: Quantum computing uses quantum bits (qubits) that can exist in superposition, representing 0 and 1 simultaneously. This allows quantum computers to process many calculations in parallel, exponentially speeding up certain problems. Key applications include cryptography, drug discovery, and optimization.

[Stream complete]

# ============================================
# AI Copywriting Assistant
# Input: product description
# Output: 3 styles of ad copy
# ============================================
from openai import OpenAI
client = OpenAI()

product = "Wireless noise-cancelling headphones, 40-hour battery, $199"

styles = [
    ("Professional", "Write a professional, feature-focused ad copy."),
    ("Emotional", "Write an emotional, lifestyle-focused ad copy."),
    ("Humorous", "Write a humorous, witty ad copy.")
]

print("=" * 60)
print(f"Product: {product}")
print("=" * 60)

for style_name, style_prompt in styles:
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[
            {"role": "system", "content": f"You are an expert copywriter. {style_prompt}"},
            {"role": "user", "content": f"Write ad copy for: {product}. Keep it under 80 words."}
        ],
        temperature=0.8,
        max_tokens=150
    )
    copy = response.choices[0].message.content
    tokens = response.usage.total_tokens
    cost = (response.usage.prompt_tokens * 0.15 + response.usage.completion_tokens * 0.60) / 1_000_000
    print(f"\n[{style_name}]")
    print(f"{copy}")
    print(f"\n(Tokens: {tokens}, Cost: ${cost:.6f})")

print("\n" + "=" * 60)

============================================================
Product: Wireless noise-cancelling headphones, 40-hour battery, $199
============================================================

[Professional]
Experience premium noise-cancelling technology with 40 hours of uninterrupted battery life. Our wireless headphones deliver crystal-clear audio and comfortable design for $199. Perfect for professionals who demand focus and quality.

(Tokens: 72, Cost: $0.000032)

[Emotional]
Escape the noise. Slip on these headphones and let the world fade away—40 hours of pure, uninterrupted bliss. Whether it's your morning commute or a quiet evening, your soundtrack awaits. Just $199 for peace you can hear.

(Tokens: 68, Cost: $0.000030)

[Humorous]
Tired of hearing your neighbor's karaoke? These noise-cancelling headphones block the chaos and deliver 40 hours of sweet silence. At $199, it's cheaper than moving. Your ears will thank you.

(Tokens: 58, Cost: $0.000025)

============================================================