language:
- en
license: mit
tags:
- mistral
- instruct
- finetune
- chatml
- gpt4
- synthetic data
- distillation
base_model: teknium/OpenHermes-2.5-Mistral-7B
model-index:
- name: MistralHermes-CodePro-7B-v1
results: []
MistralHermes-CodePro-7B-v1
In the digital pantheon of artificial intelligence, "MistralHermes-CodePro-7B-v1" stands as the architect of algorithms, a sovereign of syntax who weaves the fabric of code with unparalleled skill. This model, christened in recognition of its dual lineage—Mistral's foundational breadth and Hermes' agile conveyance—commands the binary ballet with the precision of a seasoned maestro, orchestrating the dance of data with a grace that blurs the line between the silicon and the cerebral.
Model description
MistralHermes-CodePro-7B-v1 is a fine-tuned iteration of the renowned teknium/OpenHermes-2.5-Mistral-7B model. This version has been meticulously fine-tuned using a dataset comprising over 200,000 code samples from a wide array of programming languages. It is specifically tailored to serve as a coding assistant; thus, its utility is optimized for coding-related tasks rather than a broader spectrum of applications.
Prompt Format
MistralHermes-CodePro uses the same prompt format than OpenHermes 2.5.
You should use LM Studio for chatting with the model.
Quantized Models:
GGUF: beowolx/MistralHermes-CodePro-7B-v1-GGUF
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 66.17 |
| AI2 Reasoning Challenge (25-Shot) | 62.46 |
| HellaSwag (10-Shot) | 82.68 |
| MMLU (5-Shot) | 63.44 |
| TruthfulQA (0-shot) | 49.67 |
| Winogrande (5-shot) | 77.90 |
| GSM8k (5-shot) | 60.88 |