Skip to main content

Hugging Face

Properties used to connect to Hugging Face API.

huggingFace​

Service Types​

Conversation​

  • Type: true | {
         model?: string,
         parameters?: {
             min_length?: string,
             max_length?: string,
             top_k?: string,
             top_p?: string,
             temperature?: string,
             repetition_penalty?: string},
         options?: {use_cache?: boolean}
    }

  • Default: {model: "facebook/blenderbot-400M-distill", options: {use_cache: true}}

Connect to Hugging Face Conversational API.
model is the name of the model used for the task.
min_length is the minimum length in tokens of the output summary.
max_length is the maximum length in tokens of the output summary.
top_k defines the top tokens considered within the sample operation to create new text.
top_p is a float to define the tokens that are within the sample operation of text generation. Add tokens in the sample for more probable to least probable until the sum of the probabilities is greater than top * p.
temperature is a float (ranging from 0.0 to 100.0) temperature of the sampling operation. 1 means regular sampling, 0 means always take the highest score, 100.0 is getting closer to uniform probability.
repetition_penalty is a float (ranging from 0.0 to 100.0) that controls where a token is used more within generation the more it is penalized to not be picked in successive generation passes.
use_cache is used to speed up requests by using the inference API cache.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "conversation": {"model": "facebook/blenderbot-400M-distill", "parameters": {"temperature": 1}}
    }
  }'
></deep-chat>

TextGeneration​

  • Type: true | {
         model?: string,
         parameters?: {
             top_k?: string,
             top_p?: string,
             temperature?: string,
             repetition_penalty?: string,
             max_new_tokens?: string,
             do_sample?: boolean},
         options?: {use_cache?: boolean}
    }

  • Default: {model: "gpt2", options: {use_cache: true}}

Connect to Hugging Face Text Generation API.
model is the name of the model used for the task.
top_k defines the top tokens considered within the sample operation to create new text.
top_p is a float to define the tokens that are within the sample operation of text generation. Add tokens in the sample for more probable to least probable until the sum of the probabilities is greater than top * p.
temperature is a float (ranging from 0.0 to 100.0) temperature of the sampling operation. 1 means regular sampling, 0 means always take the highest score, 100.0 is getting closer to uniform probability.
repetition_penalty is a float (ranging from 0.0 to 100.0) that controls where a token is used more within generation the more it is penalized to not be picked in successive generation passes.
max_new_tokens is an integer (ranging from 0 to 250) amount of new tokens to be generated by the response.
do_sample controls whether or not to use sampling. If false it uses greedy decoding sampling.
use_cache is used to speed up requests by using the inference API cache.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "textGeneration": {"model": "gpt2", "parameters": {"temperature": 1}}
    }
  }'
></deep-chat>

Summarization​

  • Type: true | {
         model?: string,
         parameters?: {
             min_length?: string,
             max_length?: string,
             top_k?: string,
             top_p?: string,
             temperature?: string,
             repetition_penalty?: string},
         options?: {use_cache?: boolean}
    }

  • Default: {model: "facebook/bart-large-cnn", options: {use_cache: true}}

Connect to Hugging Face Summarization API.
model is the name of the model used for the task.
min_length is the minimum length in tokens of the output summary.
max_length is the maximum length in tokens of the output summary.
top_k defines the top tokens considered within the sample operation to create new text.
top_p is a float to define the tokens that are within the sample operation of text generation. Add tokens in the sample for more probable to least probable until the sum of the probabilities is greater than top * p.
temperature is a float (ranging from 0.0 to 100.0) temperature of the sampling operation. 1 means regular sampling, 0 means always take the highest score, 100.0 is getting closer to uniform probability.
repetition_penalty is a float (ranging from 0.0 to 100.0) that controls where a token is used more within generation the more it is penalized to not be picked in successive generation passes.
use_cache is used to speed up requests by using the inference API cache.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "summarization": {"model": "facebook/bart-large-cnn", "parameters": {"temperature": 1}}
    }
  }'
></deep-chat>

Translation​

  • Type: true | {
         model?: string,
         options?: {use_cache?: boolean}
    }

  • Default: {model: "Helsinki-NLP/opus-tatoeba-en-ja", options: {use_cache: true}}

Connect to Hugging Face Translation API.
model is the name of the model used for the task.
use_cache is used to speed up requests by using the inference API cache.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "translation": {"model": "Helsinki-NLP/opus-tatoeba-en-ja"}
    }
  }'
></deep-chat>

FillMask​

  • Type: true | {
         model?: string,
         options?: {use_cache?: boolean}
    }

  • Default: {model: "bert-base-uncased", options: {use_cache: true}}

Connect to Hugging Face Fill Mask API.
model is the name of the model used for the task.
use_cache is used to speed up requests by using the inference API cache.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "fillMask": {"model": "bert-base-uncased"}
    }
  }'
></deep-chat>

QuestionAnswer​

  • Type: true | {context: string, model?: string}
  • Default: {model: "bert-large-uncased-whole-word-masking-finetuned-squad"}

Connect to Hugging Face Question Answer API.
context is a string containing details that AI can use to answer the given questions.
model is the name of the model used for the task.

Example (Ask about labrador looks)​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "questionAnswer": {
        "model": "bert-large-uncased-whole-word-masking-finetuned-squad",
        "context": "Labrador retrievers are easily recognized by their broad head, drop ears and large, expressive eyes. Two trademarks of the Lab are the thick but fairly short double coat, which is very water repellent, and the well known otter tail. The tail is thick and sturdy and comes off the topline almost straight."
        }
    }
  }'
></deep-chat>

AudioSpeechRecognition​

  • Type: true | {model?: string}
  • Default: {model: "facebook/wav2vec2-large-960h-lv60-self"}

Connect to Hugging Face Audio Speech Recognition API.
model is the name of the model used for the task.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "huggingFace": {"model": "facebook/wav2vec2-large-960h-lv60-self"}
    }
  }'
></deep-chat>

AudioClassification​

  • Type: true | {model?: string}
  • Default: {model: "ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition"}

Connect to Hugging Face Audio Classification API.
model is the name of the model used for the task.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "audioSpeechRecognition": {"model": "ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition"}
    }
  }'
></deep-chat>

ImageClassification​

  • Type: true | {model?: string}
  • Default: {model: "google/vit-base-patch16-224"}

Connect to Hugging Face Image Classification API.
model is the name of the model used for the task.

Example​

<deep-chat
  directConnection='{
    "huggingFace": {
      "key": "placeholder key",
      "imageClassification": {"model": "google/vit-base-patch16-224"}
    }
  }'
></deep-chat>