Q12 via ChatGPT: Please identify seven small language models thst have less than 80 billion parameters that were released in 2022 or 2023.

Full prompt: Please identify  seven  small language models thst have less than 80 billion parameters that were released in 2022 or 2023. Please include  Phi-2. Your responses should include each model's full name and the number of parameters it contains; it should also include the results of benchmark tests of its emergent properties.

BackToFAX

Here are seven small language models (SLMs) with less than 80 billion parameters, released in 2022 or 2023, including Phi-2:

1. Phi-2 by Microsoft:

• Parameters: 2.7 billion.
• Performance: Phi-2 showcases outstanding reasoning and language understanding capabilities. It matches or outperforms models up to 25 times larger in size on various benchmarks, demonstrating significant advancements in model scaling and training data curation​​.

2. Stable Beluga 7B:Parameters: 7 billion.

• Performance: Based on the Llama model from Meta AI, it shows robust performance across text generation, translation, question answering, and code completion, understanding and responding in multiple languages​​.

X Gen by Salesforce AI:

• Parameters: 7 billion.
• Performance: Focuses on dialogue and diverse tasks like text generation, translation, and code completion. X Gen offers computational efficiency and boasts multilingual capabilities​​.

Alibaba’s Qwen Series:

• Parameters: Models range from 1.8 billion to 72 billion.
• Performance: The series caters to diverse applications like text generation, translation, question answering, vision and language tasks, and audio processing​​.

Alpaca 7B:

• Parameters: 7 billion.
• Performance: A finely tuned replication of Meta’s LLaMA model, known for its compactness and cost-effectiveness. Alpaca 7B has demonstrated noteworthy performance, rivaling that of larger models in certain tasks​​.

MPT by Mosaic ML:

• Parameters: 7 billion.
• Performance: Designed for code generation and creative text formats, MPT excels in generating precise code snippets, automating tasks, and inspiring artistic expression through various creative text formats​​.

Falcon 7B by the Technology Innovation Institute (TII):

• Parameters: 7 billion.
• Performance: Known for its outstanding performance in tasks like chatting and question answering. Falcon 7B has been highly rated on the Hugging Face leaderboard​​.

These models represent significant strides in the field of AI and language processing, offering diverse capabilities and optimizations for various applications. They demonstrate the trend towards creating more efficient and accessible language models suitable for a wide range of applications.

BackToFAX

____________________________________

Links to related notes:  

• "The Rise of Small Language Models— Efficient & Customizable", Bijit Ghosh, Medium, 11/26/23
• "9 Best Small Language Models Released in 2023", Sandhra Jayan, AIM, 12/7/23
• "Small language models an emerging GenAI force", Antone Gonsalves, TechTarget, 12/15/23
• "Everything You Need to Know about Small Language Models (SLM) and its Applications", Tanya Malhotra, MarTechPost, 12/5/23
• "7 Steps to Running a Small Language Model on a Local CPU", Aryan Garg, KDnuggets, 11/14/23