resemble-ai/textseal | API reference

Meta TextSeal - Post-hoc text watermarking and detection via LLM rephrasing

Public

14 runs

Run resemble-ai/textseal with an API

Use one of our client libraries to get started quickly. Clicking on a library will take you to the Playground tab where you can tweak different inputs, see the results, and copy the corresponding code to use in your own project.

Input schema

The fields you can use to run this model with an API. If you don't give a value for a field its default value will be used.

Field	Type	Default value	Description
text	string		Text to watermark or detect watermark in.
mode	None	watermark	'watermark' to embed a watermark via rephrasing, 'detect' to check for an existing watermark.
watermark_type	None	gumbelmax	Watermark algorithm.
model_name	string	Qwen/Qwen2.5-1.5B-Instruct	HuggingFace model for rephrasing (watermark mode only).
secret_key	integer	42	None
temperature	number	0.9 Min: 0.1 Max: 2	None
top_p	number	0.95 Min: 0.1 Max: 1	None
delta	number	2 Min: 0.5 Max: 10	None
ngram	integer	1 Min: 1 Max: 4	None

{
  "type": "object",
  "title": "Input",
  "required": [
    "text"
  ],
  "properties": {
    "mode": {
      "enum": [
        "watermark",
        "detect"
      ],
      "type": "string",
      "title": "mode",
      "description": "'watermark' to embed a watermark via rephrasing, 'detect' to check for an existing watermark.",
      "default": "watermark",
      "x-order": 1
    },
    "text": {
      "type": "string",
      "title": "Text",
      "x-order": 0,
      "description": "Text to watermark or detect watermark in."
    },
    "delta": {
      "type": "number",
      "title": "Delta",
      "default": 2,
      "maximum": 10,
      "minimum": 0.5,
      "x-order": 7
    },
    "ngram": {
      "type": "integer",
      "title": "Ngram",
      "default": 1,
      "maximum": 4,
      "minimum": 1,
      "x-order": 8
    },
    "top_p": {
      "type": "number",
      "title": "Top P",
      "default": 0.95,
      "maximum": 1,
      "minimum": 0.1,
      "x-order": 6
    },
    "model_name": {
      "type": "string",
      "title": "Model Name",
      "default": "Qwen/Qwen2.5-1.5B-Instruct",
      "x-order": 3,
      "description": "HuggingFace model for rephrasing (watermark mode only)."
    },
    "secret_key": {
      "type": "integer",
      "title": "Secret Key",
      "default": 42,
      "x-order": 4
    },
    "temperature": {
      "type": "number",
      "title": "Temperature",
      "default": 0.9,
      "maximum": 2,
      "minimum": 0.1,
      "x-order": 5
    },
    "watermark_type": {
      "enum": [
        "greenlist",
        "gumbelmax",
        "dipmark",
        "synthid",
        "watermax"
      ],
      "type": "string",
      "title": "watermark_type",
      "description": "Watermark algorithm.",
      "default": "gumbelmax",
      "x-order": 2
    }
  }
}

Output schema

The shape of the response you’ll get when you run this model with an API.

Schema

{
  "type": "string",
  "title": "Output"
}