vicgalle/OpenHermes-Gemma-2B

Warm
Public
2.5B
BF16
8192
Feb 29, 2024
License: apache-2.0
Hugging Face
Overview

Model Overview

vicgalle/OpenHermes-Gemma-2B is a 2.5 billion parameter language model built upon the Gemma architecture. This model is designed for general-purpose conversational AI and instruction following, leveraging its compact size for efficient deployment while maintaining competitive performance on various benchmarks.

Key Capabilities

  • General Language Understanding: Achieves an average score of 46.36 on the Open LLM Leaderboard, indicating proficiency in diverse language tasks.
  • Reasoning: Scores 49.32 on the AI2 Reasoning Challenge (25-Shot) and 65.11 on Winogrande (5-shot), showcasing its ability to handle reasoning tasks.
  • Common Sense: Demonstrates common sense reasoning with a HellaSwag (10-Shot) score of 72.26.
  • Context Length: Supports an 8192-token context window, allowing for processing of moderately long inputs.

Good For

  • Efficient Inference: Its 2.5 billion parameter count makes it suitable for applications where computational resources are limited or faster inference is required.
  • Conversational Agents: Can be used as a backbone for chatbots and virtual assistants due to its general language understanding capabilities.
  • Prototyping: An excellent choice for rapid prototyping of LLM-powered features where a smaller, capable model is beneficial.