420 likes | 605 Views
ChatGPT is a large language model developed by OpenAI. It is a state-of-the-art AI system that can generate human-like responses to various queries and topics. The model is based on the transformer architecture and has been trained on a massive amount of text data. ChatGPT has several models of different sizes, from the smallest one with 124 million parameters to the largest one with 1.6 billion parameters. These models can be fine-tuned for specific tasks such as language translation, text completion, question answering, and more. With its impressive capabilities, ChatGPT has become a popular
E N D
Introduction A chatbot named ChatGPT was released by OpenAI in November 2022. It can be trained using supervised and reinforcement learning methods and is based on the Open AI GPT-3.5 families of big language models. Both methods employed human trainers to boost the model's performance. There are some restrictions on what ChatGPT can accomplish with minimal input, such as Validation Rules, Apex Code, or even a blog post. 2 www.hexaviewtech.com
Three Bases of Provided by OpenAI A set of models that can understand and generate natural language. For example: - text-DaVinci-003, text-curie-001. A set of models that can understand and generate code, including translating natural language to code. For example: - code-DaVinci-002. • A fine-tuned model that can detect whether text may be sensitive or unsafe. 3 www.hexaviewtech.com
How does it work Step 1 Collect demonstration data and train a supervised policy SFT Explain reinforcement learning to a 6 year old We give treats and punishments to teach.. A labeler demonstrates the desired output behavior. A prompt is sampled from our prompt dataset. This data is used to fine-tune GPT-3.5 with supervised learning. 4 www.hexaviewtech.com
How does it work Step 2 Collect comparison data and train a reward model. A prompt and several model outputs are sampled. A labeler ranks the outputs from best to worst. This data is used to train our reward model. Explain reinforcement learning to a 6 year old RM > > > In reinforcement learning the agent is.. > > > Explain rewards In machine learning.. We give treats and punishments to teach.. 5 www.hexaviewtech.com
How does it work Step 3 Optimize a policy against the reward model using the PPO reinforcement learning algorithm. A prompt and several model outputs are sampled. The reward model calculates a reward for the output. The PPO model is initialized from the supervised policy. The reward is used to update the policy using PPO. The policy generates an output. RM PPO r Once upon a time.. k Write a story about otters. 6 www.hexaviewtech.com
What are the implications of ChatGPT? For Cyber security ChatGPT is capable of creating viruses and phishing emails, particularly when used with OpenAI Codex. Sam Altman, the CEO of OpenAI, cautioned that new software could pose a "tremendous cybersecurity risk." Even if chatGPT was "obviously not near to AGI,“ The end of Salesforce developers A Formula, Validation Rule, Apex Class, Lightning Web Component (LWC), or Unit Test for a LWC can all be generated by ChatGPT. It can generate the XML produced by an action but not a Flow or other declarative consequence. Results from ChatGPT are textual, not graphical, making them ideal for coding. The end of Salesforce developers By posing a query and then reviewing the ChatGPT response, young developers can use ChatGPT to hone their coding abilities. The other choice is for them to pursue Prompt Engineer training. 7 www.hexaviewtech.com
Limitations Result Created by chatGPT are not Validated byMillions of websites. ChatGPT and LLMs reinforce social biases, frequently disparaging women and people of colour and fabricating historical and biographical data to support false and dangerous claims. The obvious application of ChatGPT is generating specified content, such as a Formula, rather than soliciting feedback. ChatGPT is not a replacement for experience. 8 www.hexaviewtech.com
About Us Together we foster creativity, innovation & an empowered workplace Official Blog Link https://hexaviewtech.com/blog/understanding-chatgpt-and-its-implications Transforming businesses using advanced technology by providing excellence in project, process, & product delivery and significantly impacting businesses & society around the world. Contact Us www.hexaviewtech.com +1 (646) 403-4525 marketing@hexaviewtech.com Follow Us