README.md

---
language:
- en
license: apache-2.0
library_name: transformers
tags:
- code
- QA
- reasoning
- mixtral
- maths
- sql
- mistral
- zephyr
- codellama
model-index:
- name: Moe-4x7b-reason-code-qa
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: AI2 Reasoning Challenge (25-Shot)
      type: ai2_arc
      config: ARC-Challenge
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: acc_norm
      value: 62.54
      name: normalized accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: HellaSwag (10-Shot)
      type: hellaswag
      split: validation
      args:
        num_few_shot: 10
    metrics:
    - type: acc_norm
      value: 83.87
      name: normalized accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MMLU (5-Shot)
      type: cais/mmlu
      config: all
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 61.2
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: TruthfulQA (0-shot)
      type: truthful_qa
      config: multiple_choice
      split: validation
      args:
        num_few_shot: 0
    metrics:
    - type: mc2
      value: 56.12
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Winogrande (5-shot)
      type: winogrande
      config: winogrande_xl
      split: validation
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 76.09
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: GSM8k (5-shot)
      type: gsm8k
      config: main
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 54.59
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=nextai-team/Moe-4x7b-reason-code-qa
      name: Open LLM Leaderboard
---


Model Details

Model Name: Moe-4x7b-reason-code-qa Publisher: nextai-team Model Type: Question Answering & Code Generation Architecture: Mixture of Experts (MoE) Model Size: 4x7 billion parameters

Overview

 is an advanced AI model designed by the nextai-team for the purpose of enhancing question answering and code generation capabilities. Building upon the foundation of its predecessor, Moe-4x7b-reason-code-qa, this iteration introduces refined mechanisms and expanded training datasets to deliver more precise and contextually relevant responses.

How to Use

```from transformers import AutoTokenizer
import transformers
import torch

model = "nextai-team/Moe-4x7b-reason-code-qa" 

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    device_map="auto",
    model_kwargs={"torch_dtype": torch.float16},
)

def generate_resposne(query):
    messages = [{"role": "user", "content": query}]
    prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
    return outputs[0]['generated_text']

response = generate_resposne("How to learn coding .Please provide a step by step procedure")
print(response)

```

Intended Use

This model is intended for developers, data scientists, and researchers seeking to integrate sophisticated natural language understanding and code generation functionalities into their applications. Ideal use cases include but are not limited to:

Automated coding assistance Technical support bots Educational tools for learning programming Enhancing code review processes

Model Architecture  employs a Mixture of Experts (MoE) architecture, which allows it to efficiently manage its vast number of parameters for specialized tasks. This architecture facilitates the model's ability to discern subtle nuances in programming languages and natural language queries, leading to more accurate code generation and question answering performance.

Training Data The model has been trained on a diverse and extensive corpus comprising technical documentation, open-source code repositories, Stack Overflow questions and answers, and other programming-related texts. Special attention has been given to ensure a wide range of programming languages and frameworks are represented in the training data to enhance the model's versatility.

Performance  demonstrates significant improvements in accuracy and relevance over its predecessor, particularly in complex coding scenarios and detailed technical queries. Benchmarks and performance metrics can be provided upon request.

Limitations and Biases

While  represents a leap forward in AI-assisted coding and technical Q&A, it is not without limitations. The model may exhibit biases present in its training data, and its performance can vary based on the specificity and context of the input queries. Users are encouraged to critically assess the model's output and consider it as one of several tools in the decision-making process.

Ethical Considerations

We are committed to ethical AI development and urge users to employ Moe-4x7b-reason-code-qa responsibly. This includes but is not limited to avoiding the generation of harmful or unsafe code, respecting copyright and intellectual property rights, and being mindful of privacy concerns when inputting sensitive information into the model.

Demo API: 
app.nextai.co.in

Usage Instructions

For detailed instructions on how to integrate and utilize Moe-4x7b-reason-code-qa in your projects, please refer to our GitHub repository and Hugging Face documentation.

Citation If you use Moe-4x7b-reason-code-qa in your research or application, please cite it as follows:

@misc{nextai2024moe4x7b, title={Moe-4x7b-reason-code-qa: Enhancing Question Answering and Code Generation with Mixture of Experts}, author={NextAI Team}, year={2024}, publisher={Hugging Face} }
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_nextai-team__Moe-4x7b-reason-code-qa)

|             Metric              |Value|
|---------------------------------|----:|
|Avg.                             |65.73|
|AI2 Reasoning Challenge (25-Shot)|62.54|
|HellaSwag (10-Shot)              |83.87|
|MMLU (5-Shot)                    |61.20|
|TruthfulQA (0-shot)              |56.12|
|Winogrande (5-shot)              |76.09|
|GSM8k (5-shot)                   |54.59|