Here is the curl command which only accepts one input but I need to pass a chat conversation as input in a structure like following - Human: {past-message} AI: {past-reply} Human: {new-message} Curl command for mistral instruct - curl https://api-inference.huggingface.co/models/mistralai/Mistral-…

hi @namitgupta , If you want to run a chat conversation, you need to build your input string following a chat template for the model. Here is a simple JavaScript snippet for you, I’m sure you can adapt to use with curl fetch( "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Ins…

How to Pass the Conversation as Input in the Mistral Instruct Inference API

Inference Endpoints on the Hub

radames October 12, 2023, 5:04am 4

sorry, yes you can add return_full_text to params

parameters: { max_new_tokens: 100 , return_full_text: false},

You can see all available parameters here

1 Like

Topic		Replies	Views
What is the correct format conversational models in the inference api Beginners	0	177	July 31, 2024
Using Langchain ChatHuggingface with Text Generation Inference: missing field `inputs` Beginners	2	739	August 8, 2024
Truncated output on mistralai/Mistral-7B-Instruct-v0.1 Inference Endpoints on the Hub	4	1755	December 21, 2023
Sending System and User Prompt in Deployed endpoint Beginners	3	965	November 17, 2023
Questions about Mistral and apply_chat_template with Text Generation Inference, openai API and messages API Intermediate	0	172	May 15, 2024

How to Pass the Conversation as Input in the Mistral Instruct Inference API

Related topics