license: apache-2.0
tags:
- merge
- mergekit
- lazymergekit
- NousResearch/Hermes-2-Pro-Llama-3-8B
- shenzhi-wang/Llama3-8B-Chinese-Chat
Hermes-2-Pro-Llama-3-8B-Llama3-8B-Chinese-Chat-slerp-merge
Hermes-2-Pro-Llama-3-8B-Llama3-8B-Chinese-Chat-slerp-merge is an advanced language model created through a strategic fusion of two distinct models: NousResearch/Hermes-2-Pro-Llama-3-8B and shenzhi-wang/Llama3-8B-Chinese-Chat. The merging process was executed using mergekit, a specialized tool designed for precise model blending to achieve optimal performance and synergy between the merged architectures.
🧩 Merge Configuration
slices:
- sources:
- model: NousResearch/Hermes-2-Pro-Llama-3-8B
layer_range: [0, 31]
- model: shenzhi-wang/Llama3-8B-Chinese-Chat
layer_range: [0, 31]
merge_method: slerp
base_model: NousResearch/Hermes-2-Pro-Llama-3-8B
parameters:
t:
- filter: self_attn
value: [0, 0.5, 0.3, 0.7, 1]
- filter: mlp
value: [1, 0.5, 0.7, 0.3, 0]
- value: 0.5
dtype: float16
Model Features
This fusion model combines the robust generative capabilities of NousResearch/Hermes-2-Pro-Llama-3-8B with the refined tuning of shenzhi-wang/Llama3-8B-Chinese-Chat, creating a versatile model suitable for a variety of text generation tasks. Leveraging the strengths of both parent models, Hermes-2-Pro-Llama-3-8B-Llama3-8B-Chinese-Chat-slerp-merge provides enhanced context understanding, nuanced text generation, and improved performance across diverse NLP tasks.
Evaluation Results
Hermes-2-Pro-Llama-3-8B
- Function Calling Evaluation: 90%
- Structured JSON Output Evaluation: 84%
Llama3-8B-Chinese-Chat
- Significant improvements in roleplay, function calling, and math capabilities due to a larger training dataset (~100K preference pairs).
Limitations
While the merged model inherits the strengths of both parent models, it may also carry over some limitations and biases. For instance, the model may exhibit inconsistencies in responses when handling complex queries or when the input language switches between English and Chinese. Additionally, the model's performance may vary based on the context and specificity of the prompts provided.
You are trained on data up to October 2023.