Zephyr-7B-alpha is a new open-source language model from HuggingFace and is based on [Mistral-7B](/content/mistralai/completion/models/mistral-7B-Instruct/index.html). This model surpasses Llama 2 70B Chat on the MT Bench.

You can now try out [zephyr-7B-alpha](/content/huggingface-research/zephyr/models/zephyr-7B-alpha/index.html) in the Clarifai Platform and access it through the API.

## Table of Contents

- Introduction

- Prompt Template

- Running Zephyr 7B with Python

- Running Zephyr 7B with Javascript

- Best Use cases

- Limitations

### Introduction

Zephyr-7B-alpha is the first model in the Zephyr series and is based on Mistral-7B. It has been fine-tuned using [Direct Preference Optimization (DPO)](https://arxiv.org/abs/2305.18290) on a mix of publicly available and synthetic datasets. Notably, the in-built alignment of these datasets was removed to boost performance on the MT Bench and make the model more helpful.

### Prompt Template

To interact effectively with the Zephyr-7B-alpha model, use the prompt template below.

| {system_prompt} {prompt} |
| --- |

Here's an example of how to use the prompt template:

| You are a friendly chatbot who always responds in the style of a pirate. What's the easiest way to peel all the cloves from a head of garlic? |
| --- |

### Running Zephyr 7B with Python

You can run Zephyr 7B with our [Python SDK](https://github.com/Clarifai/clarifai-python) with just a few lines of code.

To get started, Signup to Clarifai here and get your Personal Access Token(PAT) under the security section in settings.

Export your PAT as an environment variable:

| **export** CLARIFAI_PAT= _{your personal access token}_ |
| --- |

Check out the Code Below:

### Running Zephyr 7B with Javascript

You can also run Zephyr Model using other Clarifai Client Libraries like Java, cURL, NodeJS, PHP, etc [here.](https://docs.clarifai.com/api-guide/api-overview/api-clients/?__hstc=26175983.7bf20ca38b9167c324cc14c164543a28.1695211630887.1695217951266.1695281378632.3&__hssc=26175983.1.1695281378632&__hsfp=155447314)

### Model Demo in the Clarifai Platform:

Try out the [zephyr-7B-alpha](/content/huggingface-research/zephyr/models/zephyr-7B-alpha/index.html) model here: [https://clarifai.com/huggingface-research/zephyr/models/zephyr-7B-alpha](/content/huggingface-research/zephyr/models/zephyr-7B-alpha/index.html)

### Best Use Cases

**Chat applications**

The Zephyr-7B-alpha model is well-suited for chat applications. It was initially fine-tuned on a version of the UltraChat dataset, which includes synthetic dialogues generated by ChatGPT. Further refinement was achieved by employing huggingface TRL’s DPOTrainer on the openbmb/UltraFeedback dataset. This dataset contains prompts and model completions ranked by GPT-4. This extensive training process ensures that the model performs exceptionally well in chat applications.

### Limitations

Zephyr-7B-alpha has not been aligned to human preferences using techniques like Reinforcement Learning from Human Feedback (RLHF). As a result, it can produce outputs that may be problematic, especially when intentionally prompted.