Bearly provides many models and new ones are added regularly. You can choose from different models in HyperChat, regular chat and your notebooks.

HyperChat: If you click the configuration button on the left hand side of the HyperChat input field (image below), a window pops up that allows you to select different models. Please use the model dropdown to select the model you wish to use. The model selection is automatically saved.

Regular Chat: An engine Icon is on the left hand side of your chat input field. Please click the engine icon and a model selector will reveal itself allowing you to select the model you wish to use.

Notebook: An identical engine icon is available in your notebooks as well. Please select the engine icon from amongst the prompt controls and switch to your desired model. Please see the reference image for guidance.

July 25, 2023

GPT-4 and Claude V1 are our most sophisticated models. These models have many parameters, causing them to generate more polished content at a slower rate.

For speed performance and context size we recommend our ChatGPT and ClaudeV1 instant 100k models. These models have enormous context windows for large queries or long conversations. They generate quality answers swiftly, making them optimal for time sensitive tasks.

For images, StableDiffusion XL is the most advanced model. Provides high quality images with easier prompting. Slightly more time consuming than the other image models.

Whisper and Kensho Scribe offer great audio transcription. Kensho Scribe offers the additional speaker diarisation feature that allows you to see dialogue amongst the various speakers from the audio.

July 25, 2023

Token/model usage is measured via the usage meter. For the free tier the meter is always showing. For pro users, the meter will show once you surpass a 50% threshold. The meter is visible in the left panel of the main Bearly dashboard.

There are many different models that take varying inputs and produce different output types. These models affect consumption allowance at different rates. For example, 500 images may be equivalent in value to 500,000 language model tokens. Please check the billing page for the latest model usage limits.

July 25, 2023

Start by navigating to the Custom Prompt dashboard. To do this click the second icon from the top left hand side of the dashboard, labeled “My Prompts”.

Use a preconfigured template to bootstrap your prompt development from the examples dropdown. You have the option of starting fresh by clicking the “New Prompt” button in the top right. Proceed to filling out the prompt form as instructed to do so.

You will find that the most consequential field is the template field. This is where you will write your prompt template with the respective keywords that will transform into input fields that capture supplementary content. The keywords are triggered by the double bracket signature { Example }. Once you create a template field via a keyword, you will populate additional information for that field including instructions, field type, allow uploads and pasting. After you save, your prompt will be accessible via the prompt selector from the main menu. You have the option of running the prompt immediately without navigating by clicking the play button as well, the button is amongst the prompt control buttons for that particular cell/prompt.

You can invoke your custom prompt with a forward slash “/“ command in HyperChat. Then build on the prompt output via secondary chat instructions.

July 25, 2023