OpenAI-Specific Parameters in ClientAI

This guide covers the OpenAI-specific parameters that can be passed to ClientAI's generate_text and chat methods. These parameters are passed as additional keyword arguments to customize OpenAI's behavior.

generate_text Method

Basic Structure

from clientai import ClientAI

client = ClientAI('openai', api_key="your-openai-api-key")
response = client.generate_text(
    prompt="Your prompt here",          # Required
    model="gpt-3.5-turbo",             # Required
    frequency_penalty=0.5,              # OpenAI-specific
    presence_penalty=0.2,               # OpenAI-specific
    logit_bias={123: 100},             # OpenAI-specific
    max_completion_tokens=100,          # OpenAI-specific
    response_format={"type": "json"},   # OpenAI-specific
    seed=12345                         # OpenAI-specific
)

OpenAI-Specific Parameters

frequency_penalty: Optional[float]

• Range: -2.0 to 2.0
• Penalizes tokens based on their frequency in the text

  response = client.generate_text(
      prompt="Write a creative story",
      model="gpt-3.5-turbo",
      frequency_penalty=0.7  # Reduces repetition
  )
  

presence_penalty: Optional[float]

• Range: -2.0 to 2.0
• Penalizes tokens based on their presence in prior text

  response = client.generate_text(
      prompt="Write a varied story",
      model="gpt-3.5-turbo",
      presence_penalty=0.6  # Encourages topic diversity
  )
  

logit_bias: Optional[Dict[str, int]]

• Maps token IDs to bias values (-100 to 100)

  response = client.generate_text(
      prompt="Write about technology",
      model="gpt-3.5-turbo",
      logit_bias={
          123: 100,  # Increases likelihood of token 123
          456: -100  # Decreases likelihood of token 456
      }
  )
  

max_completion_tokens: Optional[int]

• Maximum tokens for completion

  response = client.generate_text(
      prompt="Write a summary",
      model="gpt-3.5-turbo",
      max_completion_tokens=100
  )
  

response_format: ResponseFormat

• Controls output structure

  response = client.generate_text(
      prompt="List three colors",
      model="gpt-4",
      response_format={"type": "json_object"}
  )
  

seed: Optional[int]

• For deterministic generation (Beta)

  response = client.generate_text(
      prompt="Generate a random number",
      model="gpt-3.5-turbo",
      seed=12345
  )
  

user: str

• Unique identifier for end-user tracking

  response = client.generate_text(
      prompt="Hello",
      model="gpt-3.5-turbo",
      user="user_123"
  )
  

chat Method

Basic Structure

response = client.chat(
    model="gpt-3.5-turbo",            # Required
    messages=[...],                    # Required
    tools=[...],                      # OpenAI-specific
    tool_choice="auto",               # OpenAI-specific
    response_format={"type": "json"}, # OpenAI-specific
    logprobs=True,                    # OpenAI-specific
    top_logprobs=5                    # OpenAI-specific
)

OpenAI-Specific Parameters

tools: Iterable[ChatCompletionToolParam]

• List of available tools (max 128)

  response = client.chat(
      model="gpt-4",
      messages=[{"role": "user", "content": "What's the weather?"}],
      tools=[{
          "type": "function",
          "function": {
              "name": "get_weather",
              "description": "Get weather data",
              "parameters": {
                  "type": "object",
                  "properties": {
                      "location": {"type": "string"}
                  }
              }
          }
      }]
  )
  

tool_choice: ChatCompletionToolChoiceOptionParam

• Controls tool selection behavior

  response = client.chat(
      model="gpt-4",
      messages=[{"role": "user", "content": "Calculate something"}],
      tool_choice="auto"  # or "none" or "required"
  )
  

modalities: Optional[List[ChatCompletionModality]]

• Output types for generation

  response = client.chat(
      model="gpt-4o-audio-preview",
      messages=[{"role": "user", "content": "Generate audio"}],
      modalities=["text", "audio"]
  )
  

audio: Optional[ChatCompletionAudioParam]

• Audio output parameters

  response = client.chat(
      model="gpt-4o-audio-preview",
      messages=[{"role": "user", "content": "Speak this"}],
      modalities=["audio"],
      audio={"model": "tts-1", "voice": "alloy"}
  )
  

metadata: Optional[Dict[str, str]]

• Custom tags for filtering

  response = client.chat(
      model="gpt-4",
      messages=[{"role": "user", "content": "Hello"}],
      metadata={"purpose": "greeting", "user_type": "new"}
  )
  

Complete Examples

Example 1: Structured Output with Tools

response = client.chat(
    model="gpt-4",
    messages=[
        {"role": "system", "content": "You are a data assistant"},
        {"role": "user", "content": "Get weather for Paris"}
    ],
    response_format={"type": "json_object"},
    tools=[{
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get weather data",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {"type": "string"}
                }
            }
        }
    }],
    tool_choice="auto"
)

Example 2: Advanced Text Generation

response = client.generate_text(
    prompt="Write a technical analysis",
    model="gpt-4",
    max_completion_tokens=500,
    frequency_penalty=0.7,
    presence_penalty=0.6,
    logit_bias={123: 50},
    user="analyst_1",
    seed=42
)

Example 3: Audio Generation

response = client.chat(
    model="gpt-4o-audio-preview",
    messages=[{"role": "user", "content": "Explain quantum physics"}],
    modalities=["text", "audio"],
    audio={
        "model": "tts-1",
        "voice": "nova",
        "speed": 1.0
    },
    metadata={"type": "educational"}
)

Parameter Validation Notes

1. Both model and prompt/messages are required
2. response_format requires compatible models
3. Tool usage limited to 128 functions
4. Audio generation requires specific models
5. logprobs must be True when using top_logprobs
6. seed feature is in Beta and not guaranteed

These parameters allow you to fully customize OpenAI's behavior while working with ClientAI's abstraction layer.