Skip to main content

Bottom Menu Bar

From Left to Right:

Nel​

Click on Nel for help. Nel is the little robot in a suit. He doesn't try to talk to you, he's just a hyperlink to the Help section.

Model Selection Menu​

Hover over the + sign to view the Model Selection menu. This is where you will select the models for your Parley.

On the left is the detail area for a Model to be added to the Active list.

Inference API​

The API profile that will be used to communicate with the provider. This is not directly editable. It will be selected based on the Template Profile you choose.

Model​

The Model from the Provider Template you have chosen. The list will be populated for most Providers. This field is directly editable if you wish to chose a model that is not yet on the pre-populated list.

  • Note: the name of the model must be exactly as expected by the Inference Provider.

System Prompt​

This message will be sent along with each message to hepl guide the model in providing a response. Most Templates will come with a generic System Prompt for your convenience. Feel free to customize it.

Inference Provider url​

This is the address of the API endpoint used to reach your Inference Provider.

  • This will be filled in for most providers.
  • Note: If you are using a custom inference address, this is where it will be entered.

Parameters​

Parameters are for tuning and define the behavior of the model, specific to the model profile.

  • The parameters must be exactly as expected by the model and Inference Provider.
  • Lines that start with a hash (#) are comments and will be ignored.

API Key​

You will need an API Key from your Inference Provider. This is like a password so it is not shown in clear text in the UI. See the specific section on Inference Providers.

Note​

The Note is just for your reference, not used in messages.

To Add Model​

Click the Activate Model button.

  • Note: Each Model profile must be unique. The profiles can be different Providers or Models, or differ in a single parameter value.

Saved​

These are model profiles you have added to the Active list at the top of the page. Select one, click the Activate Model button.

Templates​

This is where you select the Inference Provider API Template. Different Inference Providers use their own precise configuration. Most providers use the standard developed by OpenAI, with the notable exception of Google.

When a Template is chosen, there will be a list of Models to choose from for most providers. This field is editable. If the model you wish to use is not listed, enter it here.

  • Note: The model name must be exactly as the Inference Provider expects. If you are using a custom provider, this may or not be used. See the section on Inference Providers for details.

When a Saved profile is selected, you have a chance to review and edit the details before adding it to the Active List.

Prompt​

This is where you enter your question, or prompt, for the models to answer.

Submit Prompt​

Click the up arrow ( 🠉 ) to submit your prompt. The prompt will appear in a Prompt Bar in the Parley area and the models will respond.

Configuration Menu​

Hover over the hamburger ( ☰ ) to reveal the configuration menu. This is how you configure the Parley area details.

Light Mode / Dark Mode​

Light mode and dark mode available to suit your eyeballs.

Displayed Model Info​

  1. Model - The name of the model.
  2. API Template - The API Template used for this model.
  3. Note - The note entered.
  4. Provider Url - The url for this inference provider.
  5. Parameters - The specific Parameters entered for this model.
  6. System Prompt - The system prompt sent with each message.

Sticky Prompt​

Select if the Prompt Row should stay at the top of the screen for the responses while scrolling.

Vote Count​

Show or hide the message voting arrows and the vote tally in the Model info in the Active Model list.

Markdown / Monospace​

Selecting 𝑀𝒶𝓇𝓀𝒹𝑜𝓌𝓃 will show the messages formatted using Markdown. You only get fancy formatting if the model response contains Markdown.

Selecting Monospace shows the full text fo the response, specifically all the Markdown formatting symbols and characters.

Narrow Code / Wide mode​

This will have little or no effect if there are no code snippets in any of the responses.

Selecting Narrow Code will smoosh any computer code snippets so it fits into the column and maintains an orderly layout. This adds apparent line breaks and may make reviewing code more difficult, but keeps all responses in equal width columns.

Selecting Wide mode will expand the column as wide as needed to show code snippets as they are meant to be, but the columns may grow much wider and scroll horizontally off the screen. This does make copying the code to paste into your own coding environment much easier.

Download Options​

There are three choices for outputting your parley for use elsewhere.

Download Bare Table​

This is your Parley with no formatting at all. It's not meant to look pretty all by itself. Specifically used to transfer your Parley into a spreadsheet. There is no table borders, text formatting, cell coloring. Just text meant to be copied into a spreadsheet for further data mining.

  1. Select All.
  2. Copy
  3. Paste into your Spreadsheet.

Download Muted Table​

This is your Parley with minimal formatting. Lines around the columns and rows of the table, not much else.

Download Fancy Table​

This is a full copy of the Synthonnel app Parley display table. Contains all the fancy formatting. Great for full presentations.