README.md

metadata

language:
  - en
license: llama2
library_name: transformers
tags:
  - CodeMate
  - Code
  - CodeLLaMa
pipeline_tag: text-generation
model-index:
  - name: CodeMate-v0.1
    results:
      - task:
          type: text-generation
        dataset:
          name: HumanEval
          type: openai_humaneval
        metrics:
          - type: pass@1
            value: 74.9%
            name: pass@1
            verified: false
      - 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: 55.55
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          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: 78.03
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          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: 55.31
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          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: 48.64
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          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: 72.61
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          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: 40.18
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=codemateai/CodeMate-v0.1
          name: Open LLM Leaderboard

CodeMate-v0.1

CodeMate-v0.1 is an intelligent programming assistant developed by CodeMate. This model aims to assist users in generating high-quality code solutions for programming problems. Please note that this model is currently in version 0.1.

Model Details

• Training Data: Exclusively fine-tuned on a proprietary dataset of 1.8 billion tokens of high-quality programming problems and solutions.

• The dataset was generated manually and is internal to CodeMate.

• Training Techniques: The model was fine-tuned using Flash Attention 2, trained over 15 hours on 40 A100-80GB GPUs.

• A sequence length of 8096 tokens was used during training.

• Multilingual Support: CodeMate-v0.1 is proficient in multiple programming languages, including Python, C/C++, TypeScript, Java, and more.

How to Get Started with the Model

Make sure to install Transformers from the main git branch:

pip install git+https://github.com/huggingface/transformers.git

How to Prompt the Model

This model accepts prompts in the Alpaca/Vicuna instruction format. For example:

### System Prompt
You are an intelligent programming assistant.

### User Message
Implement a linked list in C++

### Assistant
...

Load the Model:

To load the model, utilize the following Python script:

from transformers import AutoTokenizer, AutoModelForCausalLM

# Initialize the model
model_path = "codemateai/CodeMate-v0.1"
model = AutoModelForCausalLM.from_pretrained(model_path, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_path)

# ... generate response ...

Bias, Risks, and Limitations

This model has undergone very limited testing. CodeMate recommends additional safety testing before any real-world deployments.

For more information and updates, visit the CodeMate website.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	58.39
AI2 Reasoning Challenge (25-Shot)	55.55
HellaSwag (10-Shot)	78.03
MMLU (5-Shot)	55.31
TruthfulQA (0-shot)	48.64
Winogrande (5-shot)	72.61
GSM8k (5-shot)	40.18