Secbone/llama-2-13B-instructed

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Sep 13, 2023License:llama2Architecture:Transformer Open Weights Warm

Secbone/llama-2-13B-instructed is a 13 billion parameter instruction-finetuned model based on the Llama 2 architecture. This model is optimized for general conversational tasks and demonstrates competitive performance across various benchmarks, including an average score of 47.84 on the Open LLM Leaderboard. It is particularly suited for applications requiring robust language understanding and generation capabilities.

Loading preview...

Secbone/llama-2-13B-instructed: An Instruction-Finetuned Llama 2 Model

This model, developed by Secbone, is an instruction-finetuned variant of the Llama 2 13B parameter model. It is designed to follow instructions effectively, making it suitable for a wide range of natural language processing tasks. The model's performance has been evaluated on the Open LLM Leaderboard, showcasing its capabilities across several key metrics.

Key Capabilities & Performance

The Secbone/llama-2-13B-instructed model achieves an average score of 47.84 on the Open LLM Leaderboard. Specific benchmark results highlight its strengths:

  • ARC (25-shot): 59.39
  • HellaSwag (10-shot): 83.88
  • MMLU (5-shot): 55.57
  • TruthfulQA (0-shot): 46.89
  • Winogrande (5-shot): 74.03

While demonstrating solid performance in general reasoning and common sense tasks, its scores on more complex reasoning benchmarks like GSM8K (8.04) and DROP (7.06) indicate areas for potential further optimization.

Good For

  • General-purpose conversational AI and chatbots.
  • Tasks requiring instruction following and text generation.
  • Applications where a 13 billion parameter model offers a balance between performance and computational resources.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p