2.2 KiB
2.2 KiB
SwarmUI API Documentation - UtilAPI
This is a subset of the API docs, see /docs/API.md for general info.
General utility API routes.
Table of Contents:
- HTTP Route CountTokens
- HTTP Route Pickle2SafeTensor
- HTTP Route TokenizeInDetail
- HTTP Route WipeMetadata
HTTP Route /API/CountTokens
Description
Count the CLIP-like tokens in a given text prompt.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| text | String | The text to tokenize. | (REQUIRED) |
| skipPromptSyntax | Boolean | If false, processing prompt syntax (things like <random:). If true, don't process that. |
False |
| tokenset | String | What tokenization set to use. | clip |
| weighting | Boolean | If true, process weighting (like (word:1.5)). If false, don't process that. |
True |
Return Format
"count": 0
HTTP Route /API/Pickle2SafeTensor
Description
Trigger bulk conversion of models from pickle format to safetensors.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| type | String | What type of model to convert, eg Stable-Diffusion, LoRA, etc. |
(REQUIRED) |
| fp16 | Boolean | If true, convert to fp16 while processing. If false, use original model's weight type. | (REQUIRED) |
Return Format
"success": true
HTTP Route /API/TokenizeInDetail
Description
Tokenize some prompt text and get thorough detail about it.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| text | String | The text to tokenize. | (REQUIRED) |
| tokenset | String | What tokenization set to use. | clip |
| weighting | Boolean | If true, process weighting (like (word:1.5)). If false, don't process that. |
True |
Return Format
"tokens":
[
{
"id": 123,
"weight": 1.0,
"text": "tok"
}
]
HTTP Route /API/WipeMetadata
Description
Trigger a mass metadata reset.
Parameters
None.
Return Format
"success": true