How to use Chat?
Chat is powered by a large language model (LLM). We always update to the latest greatest LLM's and are currently running GPT-4o. In chat, you can use the LLM to explore topics and learn new things.
- Training data: up to October 2023
- Maximum input length: 128.000 tokens (about 96.000 words)
- Maximum output length: 2.048 tokens (about 1.536 words)
Because the model is trained on information up to October 2023, it does not know about events that happened afterwards. Due to the maximum input length, prompts into the model cannot exceed 96.000 words. The maximum output length is around 1.536 words per message. This number could be increased so if you run into limitations, let us know!
On the main page of the Chat tab, you'll find options to begin a conversation. Click one of the options such as technology scouting, or technology summary to start the feature. After clicking one of the options, there is a text box at the bottom of the page where you can type your message and start the chat. Each chat option allows you to utilize a finetuned LLM to answer specific tech-related questions that match the title and subtitle of the option.
If you do not click any chat option and just start typing in the text box the ask me anything option will automatically be triggered.
For more information and examples of how to use each of the options, click the blue information icon on the righthand side of each option.
All chats will be saved automatically and are visible in the panel on the left side. Click on a chat to change the name by clicking the pencil icon, or delete the chat by clicking the bin icon.
In the upper part of the panel, there are two buttons: New Chat (to start a new chat) and Search (to search for chats by their names).
After each chat, if you are not happy with the produced results, you can edit your message using the pencil button located under the text. This will allow you to edit your original message to trigger a different response from the LLM.
After changing the message and saving it, Odin will generate a new response.
Also, there is a button for regenerating the response from Odin and another button for copying the answer from the system.
