Nous-Hermes-2-Yi-34B
| Property | Value |
|---|---|
| Parameter Count | 34.4B |
| Base Model | 01-ai/Yi-34B |
| License | Apache 2.0 |
| Training Data | 1M entries (GPT-4 generated) |
| Format | ChatML |
What is Nous-Hermes-2-Yi-34B?
Nous-Hermes-2-Yi-34B is a state-of-the-art language model that represents a significant advancement in AI capabilities. Built on the Yi-34B architecture, this model has been fine-tuned on an extensive dataset of 1,000,000 entries, primarily generated by GPT-4, along with other high-quality data sources from across the AI landscape.
Implementation Details
The model utilizes the ChatML format for interactions, making it compatible with OpenAI endpoints and familiar to ChatGPT API users. It supports system prompts for enhanced steerability and runs on BF16 precision.
- Achieves exceptional performance across multiple benchmarks including GPT4All, AGIEval, and BigBench
- Implements a sophisticated prompt format supporting multi-turn dialogue
- Available in quantized versions through GGUF format
Core Capabilities
- Strong reasoning abilities demonstrated through benchmark scores
- Excellent performance in tasks requiring logical deduction and analysis
- Enhanced truthfulness compared to previous versions (60.34% on TruthfulQA)
- Versatile dialogue capabilities with support for system-level instructions
Frequently Asked Questions
Q: What makes this model unique?
The model stands out for its significant improvements over previous Nous-Hermes versions, showing substantial gains across all benchmarks. It achieved a +5.95% average improvement over OpenHermes-2.5, with particularly strong gains in AGI Eval (+7.20%) and TruthfulQA (+7.30%).
Q: What are the recommended use cases?
The model excels in various applications including complex reasoning tasks, technical discussions, and multi-turn conversations. It's particularly well-suited for applications requiring strong logical reasoning and truthful responses, as evidenced by its benchmark performances.
