itsliupeng/llama2_7b_code
itsliupeng/llama2_7b_code is a 7 billion parameter language model based on the Llama 2 architecture, designed for general language understanding and generation tasks. With a context length of 4096 tokens, it demonstrates balanced performance across various benchmarks including ARC, HellaSwag, and MMLU. This model is suitable for applications requiring a capable general-purpose LLM within the 7B parameter class.
Loading preview...
Model Overview
itsliupeng/llama2_7b_code is a 7 billion parameter language model built upon the Llama 2 architecture, featuring a context window of 4096 tokens. This model is evaluated on the Open LLM Leaderboard, showcasing its general capabilities across a range of academic benchmarks.
Key Capabilities & Performance
The model demonstrates a balanced average performance of 42.81 on the Open LLM Leaderboard. Specific benchmark results include:
- ARC (25-shot): 52.13
- HellaSwag (10-shot): 75.71
- MMLU (5-shot): 48.05
- TruthfulQA (0-shot): 38.76
- Winogrande (5-shot): 71.51
While it shows moderate performance in reasoning tasks like ARC and MMLU, it performs well in common sense reasoning (HellaSwag, Winogrande). Its scores on GSM8K (8.11) and DROP (5.39) indicate areas for potential improvement in complex mathematical and reading comprehension tasks.
Good For
- General-purpose text generation and understanding.
- Applications requiring a Llama 2-based model in the 7B parameter range.
- Tasks benefiting from common sense reasoning and factual recall, as indicated by HellaSwag and Winogrande scores.