The CoT Collection, a new dataset created for instruction tuning, was unveiled by a research team recently. 1.88 million CoT rationales from 1,060 tasks are included in the collection. The CoT Collection dataset and the trained models are accessible through the team’s GitHub repository.

The team has carefully considered the trustworthiness, logical coherence, and informativeness of the CoT Collection in comparison to human-authored CoT rationales. The C2F2 model has also been introduced, which was developed by continuously adjusting Flan-T5 LMs with 3B and 11B parameters using the CoT Collection. It has been shown that using the CoT Collection for fine-tuning led to better zero-shot CoT performance on hidden problems.

How effectively C2F2 works in situations where learning happens in a small number of instances, or few-shot learning, is discussed in the research paper. On domain-specific datasets from the legal and medical fields, parameter-efficient fine-tuning (PEFT) on C2F2 outperforms direct fine-tuning using FLAN-T5. The benefits of utilizing CoT arguments to enhance task generalization and encourage future study have also been highlighted by the authors.

Read More: Microsoft Announces AI Personal Assistant Windows Copilot for Windows 11

In order to determine the degree of improvement following the use of the CoT Collection, the researchers assessed the average zero-shot accuracy on 27 datasets of the BIG-Bench-Hard benchmark. The 3B and 11B LMs’ accuracy improved by +4.34% and +2.44%, respectively. The few-shot learning capabilities of the language models were also enhanced by the CoT instruction modification. This resulted in improvements of +2.97% and +2.37% on four domain-specific tasks as compared to Flan-T5 LMs (3B and 11B), respectively.

In comparison to earlier CoT datasets, the CoT Collection contains over 52 times as many CoT justifications and roughly 177 times as many jobs. The CoT Collection dataset, in conclusion, demonstrates the efficacy of CoT justifications for enhancing task generalization in Language Models under zero-shot and few-shot learning conditions. It overcomes the difficulties encountered when applying CoT reasoning in more compact language models.

In an effort to promote continuous learning within the company, Air India has introduced a cutting-edge learning hub. Known as Gurukul.AI, the hub has been created with Vihaan.AI. 

Through an evaluation of each employee’s job functions, existing abilities, and proficiencies, the airline’s five-year transformation plan intends to establish customized upskilling paths for each employee. According to the airline, the platform incorporates competency frameworks that are tied to key organizational roles and enables access to pertinent courses.

According to the platform’s description, its main goal is to cultivate state-of-the-art, world-class capabilities within Air India, improving employee productivity and skill sets to the highest standards possible. Emerging technologies within the portal will allow employees to view their progress via automated analytics and assist them in performance management. 

Read More: Microsoft Announces AI Personal Assistant Windows Copilot for Windows 11

In keeping with it, the platform includes game-like components and hyper-personalization, such as a function that may “talk” to the students. The airline said that doing this would inspire staff to advance, reach milestones, and unlock achievements, ultimately fostering a sense of accomplishment.

In addition, Gurukul.AI offers a collection of more than 70,000 cutting-edge learning resources, such as microlearning, mobile learning resources that are readily accessible, and modules that are engaging and video-based. 

The organization believes that features like a leader board built into the platform, along with a learning wallet and chances to earn rewards, would encourage active engagement in team learning and serve as an incentive for employees to improve their knowledge and skills.

UpCodes is unveiling a new AI-based tool Copilot that will further streamline the world of writing code. Copilot, based on GPT-4, serves as a research assistant by responding to complex code questions and providing explanations with links to pertinent code sections.

UpCodes’ primary goal prior to the release of Copilot was to create its own database of codes, digitize regulations that were frequently only available in printed reference materials, and make them simple to examine. It contains 160,000 local modifications in addition to five million code parts. Since codes are continually changing, UpCodes releases an average of 7,000 changes per month.

Copilot contains a searchable database and additional tools, like its code check feature, that are intended to make code compliance simpler, but because of the complicated regulations, these are also difficult to use at first. Copilot wants to make the process of finding new codes incredibly simple. Copilot responds to these inquiries and assists users by listing the code source from which it derives the same so they can examine the actual code for themselves.

Read More: Microsoft Announces AI Personal Assistant Windows Copilot for Windows 11

For more than seven years, UpCodes has helped both professionals in the industry and amateurs better grasp the complicated world of building codes. On its website, you may explore a library of laws from every state and take advantage of tools like a “spell check” that highlights coding mistakes. 

Additionally, Upcodes disclosed that it has closed a $3.5 million Series A round of funding with the intention of developing Copilot and expanding the platform’s AI-based features. UpCodes has now raised $7.6 million in total, including prior funding and a Pre-Series A that was announced in March 2021.

Building Ventures, a VC firm specializing in construction and real estate tech, took the lead in the most recent round. CapitalX, Bragiel Bros., and the co-founders of PlanGrid are other participants.