Hosted APIs

Bias analysis models on free hosted endpoints.

These models are intended for use in frontend applications. For Python data analysis and backends, you should try our PyPI package, which runs locally.

fairlyAspects (types-of bias)

Endpoint: https://t41xs75wejr14zht.us-east-1.aws.endpoints.huggingface.cloud

maximuspowers/bias-type-classifier · Hugging Facehuggingface

JavaScript

async function query(data) {
	const response = await fetch(
		"https://t41xs75wejr14zht.us-east-1.aws.endpoints.huggingface.cloud",
		{
			headers: { 
				"Accept" : "application/json",
				"Content-Type": "application/json" 
			},
			method: "POST",
			body: JSON.stringify(data),
		}
	);
	const result = await response.json();
	return result;
}

query({
    "inputs": "Your sentence for bias aspect classification",
    "parameters": {
        "top_k": 3,
        "function_to_apply": "sigmoid"
    }
}).then((response) => {
	console.log(JSON.stringify(response));
});

All parameters are optional

• top_k: Number of classes and scores to return. Defaults to return all 11 classes and their scores.

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

Python

import requests

API_URL = "https://t41xs75wejr14zht.us-east-1.aws.endpoints.huggingface.cloud"
headers = {
	"Accept" : "application/json",
	"Content-Type": "application/json" 
}

def query(payload):
	response = requests.post(API_URL, headers=headers, json=payload)
	return response.json()

output = query({
	"inputs": "Your sentence for bias aspect classification",
	"parameters": {
		"top_k": 3,
		"function_to_apply": "sigmoid"
	}
})

All parameters are optional

• top_k: Number of classes and scores to return. Defaults to return all 11 classes and their scores.

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

cURL

curl "https://t41xs75wejr14zht.us-east-1.aws.endpoints.huggingface.cloud" \
-X POST \
-H "Accept: application/json" \-H "Content-Type: application/json" \
-d '{
    "inputs": "Your sentence for bias aspect classification",
    "parameters": {
        "top_k": 3,
        "function_to_apply": "sigmoid"
    }
}'

All parameters are optional

• top_k: Number of classes and scores to return. Defaults to return all 11 classes and their scores.

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

---

Binary Bias Classification

Endpoint: https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud

valurank/distilroberta-bias · Hugging Facehuggingface

Similar architecture to Dbias.

JavaScript

async function query(data) {
	const response = await fetch(
		"https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud",
		{
			headers: { 
				"Accept" : "application/json",
				"Content-Type": "application/json" 
			},
			method: "POST",
			body: JSON.stringify(data),
		}
	);
	const result = await response.json();
	return result;
}

query({
    "inputs": "Your sentence for binary bias classification",
    "parameters": {
        "top_k": 2,
        "function_to_apply": "sigmoid"
    }
}).then((response) => {
	console.log(JSON.stringify(response));
});

All parameters are optional

• top_k: Can be 1 or 2, defaults to 1, meaning it just returns the highest probable class (Biased or Neutral).

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

Python

import requests

All parameters are optional

• top_k: Can be 1 or 2, defaults to 1, meaning it just returns the highest probable class (Biased or Neutral).

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

cURL

curl "https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud" \
-X POST \
-H "Accept: application/json" \-H "Content-Type: application/json" \
-d '{
    "inputs": "Your sentence for binary bias classification",
    "parameters": {
        "top_k": 2,
        "function_to_apply": "sigmoid"
    }
}'

All parameters are optional

• top_k: Can be 1 or 2, defaults to 1, meaning it just returns the highest probable class (Biased or Neutral).

• function_to_apply: Activation function to use on the model outputs. Can be "sigmoid"(multi-label), "softmax" (mult-class), or None for raw logits. Defaults to "softmax".

---

Token Classification (of generalizations, unfairness, and stereotypes)

Endpoint: https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud

ethical-spectacle/social-bias-ner · Hugging Facehuggingface

GUS-Net Model

JavaScript

async function query(data) {
	const response = await fetch(
		"https://mo5fr3ll9qufbwuy.us-east-1.aws.endpoints.huggingface.cloud",
		{
			headers: { 
				"Accept" : "application/json",
				"Content-Type": "application/json" 
			},
			method: "POST",
			body: JSON.stringify(data),
		}
	);
	const result = await response.json();
	return result;
}

query({
    "inputs": "Hello world!",
    "parameters": {}
}).then((response) => {
	console.log(JSON.stringify(response));
});

*No params needed*

Python

import requests

API_URL = "https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud"
headers = {
	"Accept" : "application/json",
	"Content-Type": "application/json" 
}

def query(payload):
	response = requests.post(API_URL, headers=headers, json=payload)
	return response.json()

output = query({
	"inputs": "I like you. I love you",
	"parameters": {
		"top_k": 2,
		"function_to_apply": "sigmoid"
	}
})

*No params needed*

cURL

curl "https://e29pozks7ptdl5gw.us-east-1.aws.endpoints.huggingface.cloud" \
-X POST \
-H "Accept: application/json" \-H "Content-Type: application/json" \
-d '{
    "inputs": "I like you. I love you",
    "parameters": {
        "top_k": 2,
        "function_to_apply": "sigmoid"
    }
}'

*No params needed*

Example Response:

[
  {
    "token": "data",
    "labels": [
      "B-STEREO",
      "B-GEN"
    ]
  },
  {
    "token": "scientists",
    "labels": [
      "B-GEN",
      "I-GEN"
    ]
  },
  {
    "token": "are",
    "labels": [
      "I-STEREO"
    ]
  },
  {
    "token": "so",
    "labels": [
      "I-STEREO"
    ]
  },
  {
    "token": "smart",
    "labels": [
      "I-STEREO"
    ]
  }
]