GenerateRequest

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "#/components/schemas/GenerateRequest",
  "title": "GenerateRequest",
  "type": "object",
  "required": [
    "inputs"
  ],
  "properties": {
    "inputs": {
      "type": "string",
      "description": "Input prompt text",
      "example": "What is deep learning?"
    },
    "parameters": {
      "type": "object",
      "properties": {
        "max_new_tokens": {
          "type": "integer",
          "description": "Maximum number of tokens to generate",
          "default": 20,
          "example": 100
        },
        "temperature": {
          "type": "number",
          "format": "float",
          "description": "Sampling temperature (higher = more random)",
          "minimum": 0,
          "example": 0.7
        },
        "top_p": {
          "type": "number",
          "format": "float",
          "description": "Nucleus sampling probability threshold",
          "minimum": 0,
          "maximum": 1,
          "example": 0.95
        },
        "top_k": {
          "type": "integer",
          "description": "Top-k sampling parameter",
          "minimum": 0,
          "example": 50
        },
        "repetition_penalty": {
          "type": "number",
          "format": "float",
          "description": "Repetition penalty (1.0 = no penalty)",
          "example": 1.1
        },
        "do_sample": {
          "type": "boolean",
          "description": "Whether to use sampling vs greedy decoding",
          "default": false
        },
        "seed": {
          "type": "integer",
          "description": "Random seed for reproducibility"
        },
        "stop": {
          "type": "array",
          "items": {
            "type": "string"
          },
          "description": "Stop sequences"
        },
        "watermark": {
          "type": "boolean",
          "description": "Whether to add a watermark to generated text",
          "default": false
        },
        "return_full_text": {
          "type": "boolean",
          "description": "Include the input prompt in the response",
          "default": false
        },
        "decoder_input_details": {
          "type": "boolean",
          "description": "Return decoder input token details",
          "default": false
        },
        "details": {
          "type": "boolean",
          "description": "Return generation details (tokens, logprobs, etc.)",
          "default": false
        },
        "truncate": {
          "type": "integer",
          "description": "Truncate input to this many tokens"
        },
        "typical_p": {
          "type": "number",
          "format": "float",
          "description": "Typical decoding probability threshold"
        },
        "best_of": {
          "type": "integer",
          "description": "Generate this many sequences and return the best",
          "minimum": 1
        },
        "grammar": {
          "type": "object",
          "description": "Grammar constraints for generation",
          "properties": {
            "type": {
              "type": "string",
              "enum": [
                "json",
                "regex"
              ]
            },
            "value": {
              "type": "string",
              "description": "Grammar specification (JSON schema or regex)"
            }
          }
        }
      },
      "example": "example_value"
    }
  }
}