YeonwooSung/Bloslain-8B-v0.2
YeonwooSung/Bloslain-8B-v0.2 is an 8 billion parameter merged language model based on the Llama 3.1 architecture, created by YeonwooSung. This model combines several Llama 3.1 variants, including those focused on Python coding and reflection, to enhance its overall capabilities. With a context length of 32768 tokens, it is designed for general language tasks with specialized strengths derived from its merged components.
Loading preview...
Bloslain-8B-v0.2 Overview
Bloslain-8B-v0.2 is an 8 billion parameter language model developed by YeonwooSung, built upon the Llama 3.1 architecture. This model is a result of merging three distinct Llama 3.1-based models using the LazyMergekit tool: BlackBeenie/Neos-Llama-3.1-8B, Solshine/Meta-Llama-3.1-8B-Instruct-Python-Coder, and Solshine/reflection-llama-3.1-8B. The merge method employed is dare_ties, with specific density and weight parameters applied to each contributing model.
Key Capabilities
- Enhanced General Performance: By combining multiple Llama 3.1 variants, Bloslain-8B-v0.2 aims for improved performance across a range of tasks.
- Python Coding Focus: The inclusion of a Python Coder model suggests a specialized capability in understanding and generating Python code.
- Reflection Abilities: Integration of a reflection-focused model indicates potential for improved reasoning and self-correction in responses.
Performance Metrics
Evaluations on the Open LLM Leaderboard show an average score of 23.80. Specific benchmark results include:
- IFEval (0-Shot): 50.23
- BBH (3-Shot): 30.66
- MATH Lvl 5 (4-Shot): 14.50
- MMLU-PRO (5-shot): 29.48
Good For
- Applications requiring a general-purpose LLM with a Llama 3.1 foundation.
- Tasks benefiting from enhanced Python code generation or comprehension.
- Use cases where improved reasoning or 'reflection' capabilities are advantageous.