uukuguy/speechless-llama2-luban-orca-platypus-13b
The uukuguy/speechless-llama2-luban-orca-platypus-13b is a 13 billion parameter language model, merged from AIDC-ai-business/Luban-13B and Open-Orca/OpenOrca-Platypus2-13B, built upon the Llama 2 architecture with a 4096-token context length. This model demonstrates strong performance across various academic benchmarks, including ARC, HellaSwag, MMLU, and TruthfulQA, making it suitable for general-purpose natural language generation and understanding tasks. Its merged origin suggests a focus on combining the strengths of its constituent models for enhanced reasoning and conversational capabilities.
Loading preview...
Model Overview
uukuguy/speechless-llama2-luban-orca-platypus-13b is a 13 billion parameter language model derived from a merge of AIDC-ai-business/Luban-13B and Open-Orca/OpenOrca-Platypus2-13B. It leverages the Llama 2 architecture, known for its optimized transformer design and a 4096-token context window. The model's development aimed to combine the strengths of its merged components, resulting in a versatile LLM.
Key Capabilities & Performance
This model exhibits solid performance across a range of academic benchmarks, indicating its proficiency in various NLP tasks:
- ARC: 62.54
- HellaSwag: 82.76
- MMLU: 59.23
- TruthfulQA: 54.66
These scores suggest strong capabilities in commonsense reasoning, factual recall, and general language understanding. The underlying Llama 2 models were trained on 2 trillion tokens of publicly available data, with fine-tuning incorporating over a million human-annotated examples, enhancing its ability to align with human preferences.
Intended Use Cases
- General-purpose text generation: Suitable for a wide array of natural language generation tasks.
- Research and commercial applications: Designed for use in English-language contexts.
- Assistant-like chat: While the base Llama 2 models have chat-optimized variants, this merged model likely benefits from the conversational fine-tuning of its components.
Limitations
As with all LLMs, this model carries inherent risks. It has been primarily tested in English, and its outputs may occasionally be inaccurate, biased, or objectionable. Developers are advised to conduct thorough safety testing for specific applications.