HomeIndustriesMeet “Smaug-72B”: the brand new king of open source AI

Meet “Smaug-72B”: the brand new king of open source AI

Accordingly, a brand new open source language model has conquered the throne of the very best on this planet Latest rankings from Hugging Faceone among the leading platforms for natural language processing (NLP) research and applications.

The model called “Smaug-72B“, was released publicly by the startup today Abacus AI, which helps firms solve difficult problems in artificial intelligence and machine learning. Smaug-72B is technically a fine-tuned version of “Qwen-72B“, one other powerful language model released just a number of months ago by Qwen, a team of researchers at Alibaba Group.

What's most notable about today's release is that Smaug-72B outperforms GPT-3.5 and Mistral Medium, two of essentially the most advanced proprietary large language models developed by OpenAI and Mistral, respectively, in several of the preferred benchmarks. Smaug-72B also significantly outperforms Qwen-72B, the model from which it was derived, in a lot of these assessments.

According to the Hugging Face Open LLM RankingsSmaug-72B, which measures the performance of open source language models on quite a lot of natural language understanding and generation tasks, is now the primary and only open source model to realize a mean rating of over in all major LLM assessments than 80.

While the model still doesn't reach the 90-100 point average that indicates human-level performance, its emergence signals that open-source AI could soon rival the capabilities of Big Tech, long kept secret were wrapped. In short, the discharge of Smaug-72B could fundamentally change the best way AI advances unfold, harnessing the ingenuity of those beyond only a handful of rich firms.

The open source advantage

“Smaug-72B by Abacus AI is now available on Hugging Face, running is at the highest of the LLM rankings and is the primary model with a mean rating of 80!! In other words, it’s the world’s best open source foundation model,” Abacus AI CEO Bindu Reddy said in a post on X.com.

“Our next goal shall be to publish these techniques as a research paper and apply them to a few of the very best Mistral models, including Miqu (a 70B effective tine from LLama-2),” she added. “The techniques we use specifically goal pondering and math skills, which explains the high GSM8K values! Our upcoming paper will explain more.”

With today's release, Smaug-72B becomes the primary open source model to realize a mean rating of 80 on the Hugging Face Open LLM leaderboard, considered a notable achievement in the sphere of natural language processing and open source AI.

Thanks to the techniques Abacus AI utilized in fine-tuning, Smaug-72B particularly excels at pondering and math tasks. These techniques are described in additional detail in a upcoming research workgoal the weaknesses of huge language models and improve their capabilities.

Smaug-72B isn't the one open source language model to make headlines recently. Qwen, the group behind Qwen-72B, too released Qwen 1.5a set of small, high-performance language models with parameters from 0.5B to 72B.

Qwen 1.5 outperforms popular proprietary models corresponding to Mistral-Medium and GPT-3.5, has a context length of 32,000, and works with various tools and platforms for fast and native inference. Qwen can also be open source Qwen-VL-Maxa brand new large vision language model that competes with Gemini Ultra and GPT-4V, two of essentially the most advanced proprietary vision language models developed by Google and OpenAI, respectively.

Implications for the long run of AI

The creation of Smaug-72B And Qwen 1.5 has sparked much excitement and debate within the AI ​​community and beyond. Many experts and influencers have praised the achievements of Abacus AI and Qwen and expressed their admiration for his or her contribution to open source AI.

“It's hard to consider that lower than a yr ago we were all enthusiastic about models like Dolly,” said Sahar Mor, an AI influencer and analyst. in a LinkedIn postwho’s pleased with the progress made by open source models up to now yr.

Smaug-72B and Qwen 1.5 are currently available on Hugging Face, where anyone can download, use and modify them. Abacus AI and Qwen have also announced their plans to submit their models to the llmsys human rating leaderboardThis is a brand new benchmark that evaluates the performance of language models in human-like tasks and scenarios. Abacus AI and Qwen have also hinted at their future projects and goals, which include creating more open source models and applying them to different domains and applications.

Smaug-72B and Qwen 1.5 are only the most recent examples of the rapid and noteworthy development of open source AI this yr. They represent a brand new wave of AI innovation and democratization that challenges the dominance and monopoly power of massive tech firms and opens up latest possibilities and opportunities for everybody. Only time will tell how long Smaug-72B will stay at the highest of the Hugging Face leaderboard, but for now it's protected to say that open source AI is having an enormous moment early within the yr.


Please enter your comment!
Please enter your name here

Must Read