A recent advancement in the field of artificial intelligence is the introduction of large scale language models (LLMs). These models make understanding language more concise and allow you to take full advantage of natural language processing (NLP) and natural language understanding (NLU). These models perform well on all other tasks such as text summarization, question answering, content generation, and language translation. They understand complex textual prompts containing reasoning and logic, and identify patterns and relationships in that data.
Although language models have come a long way in recent years by showing tremendous performance and demonstrating their abilities on a wide variety of tasks, efficient use of the tools via API calls remains challenging. Even well-known LLMs like GPT-4 struggle to generate accurate input arguments and often recommend bad API calls. To address this problem, researchers at Berkeley and Microsoft Research proposed Gorilla, a tweaked version of his LLaMA-based model that outperforms his GPT-4 in generating API calls. Gorilla helps you choose the right APIs and improves your LLM’s ability to work with external tools to perform specific activities.
The research team also created the APIBench dataset, which consists of a large corpus of APIs with overlapping functionality. This dataset was created by collecting his TorchHub, TensorHub, HuggingFace and other public model hubs for the ML API. All API requests from TorchHub and TensorHub are included per API, and the top 20 models from HuggingFace are selected for each task category. In addition, we generate 10 hypothetical user query prompts for each API using a self-directing method.
Using this APIBench dataset and document search, researchers fine-tuned Gorilla. Gorilla’s 7 billion parameter model outperforms GPT-4 in terms of API function accuracy and reduces hallucinogenic errors. The effective integration of the Document Retrieval Tool with Gorilla demonstrates the potential for LLMs to use the tool more accurately. Gorilla’s improved ability to generate API calls and the ability to modify the documentation as needed increases the applicability and reliability of model results. This development is important because it allows LLMs to stay on top of regularly updated documentation and provide users with more accurate and up-to-date information.
One example shared by researchers shows how Gorilla correctly recognizes tasks and provides fully qualified API results. The API calls generated by the model show GPT-4 generating API requests to the hypothetical model, demonstrating a lack of understanding of the task. Claude chose the wrong library, demonstrating his lack of ability to recognize the appropriate resource. In contrast, gorillas correctly perceive the task. Gorilla therefore differs from GPT-4 and Claude as his API call making is precise and demonstrates improved performance and task comprehension.
In conclusion, Gorilla adds significantly to the list of language models as it also addresses the issue of making API calls. That feature can help alleviate problems related to hallucinations and credibility.
Please check Paper, Github linkand project page. don’t forget to join 22,000+ ML SubReddit, Discord channeland email newsletterShare the latest AI research news, cool AI projects, and more. If you have any questions regarding the article above or missed something, feel free to email me. Asif@marktechpost.com
🚀 Check out 100’s of AI Tools at the AI Tools Club
Tanya Malhotra is a final year student at the University of Petroleum and Energy Research, Dehradun, graduating with a Bachelor of Science in Computer Science Engineering with a specialization in Artificial Intelligence and Machine Learning.
A data science enthusiast with good analytical and critical thinking, she has a keen interest in learning new skills, leading groups, and managing work in an organized manner.
