arcee-ai/Hermes-Mistral-Legal-Slerp

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kTool Calling:SupportedPublished:Mar 7, 2024License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Hermes-Mistral-Legal-Slerp is a 7 billion parameter language model created by arcee-ai, formed by merging NousResearch/Nous-Hermes-2-Mistral-7B-DPO and mistralai/Mistral-7B-v0.1+predibase/legal using the slerp method. This model combines general instruction-following capabilities with specialized legal domain knowledge. It is designed for applications requiring both broad understanding and specific legal reasoning, operating within a 4096 token context length.

Loading preview...

Hermes-Mistral-Legal-Slerp Overview

Hermes-Mistral-Legal-Slerp is a 7 billion parameter language model developed by arcee-ai. It is a merged model, combining the strengths of two distinct base models: NousResearch's Nous-Hermes-2-Mistral-7B-DPO and mistralai/Mistral-7B-v0.1+predibase/legal. The merge was performed using the slerp (spherical linear interpolation) method via mergekit.

Key Capabilities

  • Hybrid Expertise: Integrates the general instruction-following and conversational abilities of Nous-Hermes-2-Mistral-7B-DPO with the specialized legal domain understanding from the predibase/legal model.
  • Merged Architecture: Leverages the Mistral-7B architecture, known for its efficiency and performance in its size class.
  • Configurable Merge: The merge process utilized specific t parameters for self-attention and MLP layers, indicating a fine-tuned balance between the contributing models' characteristics.

Good For

  • Legal Applications: Ideal for tasks requiring an understanding of legal texts, terminology, and concepts, such as legal research, document analysis, or generating legally-informed responses.
  • Instruction Following: Benefits from the DPO-tuned Nous-Hermes base, making it suitable for a wide range of instruction-based tasks beyond just the legal domain.
  • Resource-Efficient Deployment: As a 7B parameter model, it offers a balance of capability and computational efficiency, making it viable for deployment in environments with moderate resource constraints.