ServerlessLLMChat
Mixtral MoE 8x7B Instruct (HF Version) is the original, FP16 version of Mixtral MoE 8x7B Instruct whose results should be consistent with the official Hugging Face implementation.
Mixtral MoE 8x7B Instruct (HF version) is available via Fireworks' serverless API, where you pay per token. There are several ways to call the Fireworks API, including Fireworks' Python client, the REST API, or OpenAI's Python client.
See below for easy generation of calls and a description of the raw REST API for making API requests. See the Querying text models docs for details.
Generate a model response for a given chat conversation. API reference
import requests import json url = "https://api.fireworks.ai/inference/v1/chat/completions" payload = { "model": "accounts/fireworks/models/mixtral-8x7b-instruct-hf", "max_tokens": 4096, "top_p": 1, "top_k": 40, "presence_penalty": 0, "frequency_penalty": 0, "temperature": 0.6, "messages": [] } headers = { "Accept": "application/json", "Content-Type": "application/json", "Authorization": "Bearer <API_KEY>" } requests.request("POST", url, headers=headers, data=json.dumps(payload))
Mixtral MoE 8x7B Instruct (HF version) can be fine-tuned on your data to create a model with better response quality. Fireworks uses low-rank adaptation (LoRA) to train a model that can be served efficiently at inference time.
See the Fine-tuning guide for details.
On-demand deployments allow you to use Mixtral MoE 8x7B Instruct (HF version) on dedicated GPUs with Fireworks' high-performance serving stack with high reliability and no rate limits.
See the On-demand deployments guide for details.