The launch of a new language generation AI model called “Grok” by a company, which is claimed to be more capable than GPT-3.5. Elon Musk announced this model’s launch on Saturday, mentioning it was developed after “The Hitchhiker’s Guide to the Galaxy,” intending to provide answers to almost any query and even suggest what questions should be asked.
Grok will be included in ‘X’, previously known as Twitter, designed with a humorous approach to answering questions. It’s specifically advised that if someone dislikes humor, they shouldn’t use Grok.
This AI aims to understand humanity and aid in acquiring knowledge. It operates on Grok-1 GPT-3.5, developed within a four-month period. The prototype Grok-0 was trained with 33 billion parameters, considered as competent as Meta’s LLaMA 2, supporting 70 billion parameters.
According to benchmarks, Grok-1 achieves 63.2% on human-like coding tasks and 73% on math-related tasks compared to GPT-4. Within a limited time, the company claims they can improve Grok-1’s performance compared to Grok-0.
In benchmarks related to mathematical word problems (GSM8k), Grok-1 scored 62.9%, surpassing GPT-3.5 and LLaMa 2 but falling short of PaLM 2, Cloud 2, and GPT-4.
These numbers indicate that Grok-1 is more capable than OpenAI’s GPT-3.5 but not as capable as the latest model, GPT-4. The company claims that despite being trained on less data, Grok-1 can outperform models trained on larger datasets that require high computing capabilities.
Grok-1 was trained using a custom training and inference stack based on Kubernetes, Rust, and JAX. As it has real-time access to the latest information and the internet, the company claims it can generate incorrect or biased information.
To address these issues in future models, xAI is seeking human feedback, contextual understanding, multimodal capabilities, and robustness.
Currently, the beta version of Grok is available to a limited number of users in the US. In the coming days, it will be made available to X Premium+ customers at a monthly cost of 1,300 rupees upon subscription through the desktop.
Grok’s capabilities are continually evolving, with the company focusing on mitigating potential issues by incorporating more human-centric approaches. Acknowledging the importance of ethical considerations, the company is exploring how Grok can responsibly handle sensitive and contentious topics, ensuring that the information provided is reliable and unbiased.
The development roadmap for Grok includes a dedication to enhancing its interpretability and explainability. The aim is to make the AI more transparent in how it arrives at its conclusions or generates responses. This increased transparency will not only boost user confidence but also aid in better understanding the AI’s decision-making process.
To get the Early access of the GROK AI click https://grok.x.ai/
Furthermore, Grok is being tested across various domains to ensure its adaptability and accuracy in different scenarios. The company is committed to continually improving and fine-tuning the model based on user feedback and emerging technological advancements. Grok’s expansion to a broader audience in the near future is anticipated, aiming to revolutionize access to advanced AI capabilities for individuals across diverse fields.
For more latest news click https://fastnewspoint.com/