Skip to content

Commit 3f0b7a0

Browse files
MuyangAmigontrogh
andauthored
Refresh AITK Playground Page (#8465)
* Update playground doc * Update playground doc with new contents * Perform doc review * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update docs/intelligentapps/playground.md Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com> * Update for comments and remove Foundry part --------- Co-authored-by: Nick Trogh <1908215+ntrogh@users.noreply.github.com>
1 parent c371a76 commit 3f0b7a0

File tree

9 files changed

+106
-22
lines changed

9 files changed

+106
-22
lines changed
Lines changed: 3 additions & 0 deletions
Loading
Lines changed: 3 additions & 0 deletions
Loading
Lines changed: 3 additions & 0 deletions
Loading
Lines changed: 3 additions & 0 deletions
Loading
Lines changed: 2 additions & 2 deletions
Loading
Lines changed: 2 additions & 2 deletions
Loading
Lines changed: 3 additions & 0 deletions
Loading
Lines changed: 3 additions & 0 deletions
Loading

docs/intelligentapps/playground.md

Lines changed: 84 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -1,40 +1,106 @@
11
---
22
ContentId: e919aee8-fd2e-401b-9d83-0ff6f98b23ba
3-
DateApproved: 12/11/2024
3+
DateApproved: 14/06/2025
44
MetaDescription: Chat with selected generative AI model in playground. Change system prompt and parameters. Add attachment for Multi-Modal models. Keep chat history.
55
---
6-
# AI Toolkit playground
6+
# Model playground in AI Toolkit
7+
The AI Toolkit playground provides an interactive environment to experiment with generative AI models. You can test various prompts, adjust model parameters, compare responses from different models and explore multi-modal capabilities by attaching different types of input files.
78

8-
The AI Toolkit playground enables you to interact with your AI models and try different prompts with different model parameter settings. You can also use the playground to interact with multi-modal models that support attachment of different input formats.
9-
10-
![Playground view](./images/playground/playground.png)
9+
![AI Toolkit playground interface showing a chat interaction with a generative AI model.](./images/playground/playground.png)
1110

1211
## Test a model in the playground
1312

1413
To access the playground:
1514

16-
- In AI Toolkit view, select **Playground**
17-
18-
- Select **Load in Playground** or **Try in Playground** from a model card in the model catalog
15+
- In AI Toolkit view, select **Tools** > **Playground**
16+
- Select **Try in Playground** from a model card in the model catalog
17+
- Double click a model from **MY Models** to open it in the playground
1918

2019
To test a model in the playground, follow these steps:
2120

22-
1. In **Model Preferences**, select a model from the dropdown list
21+
1. In **Model Preferences**, select a model from the dropdown list.
22+
2. Optionally, add a **System Prompt** to guide the model response.
23+
3. Optionally, configure the model parameters available for the selected model.
24+
4. Enter a chat prompt in the chat input box
25+
26+
## Model preferences
27+
![Model Preferences panel displaying adjustable parameters for generative AI models](./images/playground/parameters.png)
28+
29+
The **Model Preferences** panel allows you to select a model for playground interaction, set system prompts to guide the model's behavior and configure the model parameters for the selected model.
30+
31+
### Select a model
32+
The dropdown list in the **Model Preferences** panel shows the models you have added to AI Toolkit. You can select **Browse Models** to open the model catalog and add more models.
33+
34+
### System prompt
35+
A system prompt is a predefined instruction that guides the model's behavior and response style. You can add, edit, or delete the system prompt in the input text area to influence how the model generates responses.
36+
37+
### Configure inference parameters
38+
The available parameters depend on the model type and publisher. Common parameters include:
39+
- Temperature: Controls the randomness of the model's output. Lower values make the output more deterministic. Higher values introduce more variability.
40+
- Top P: Controls the diversity of the output by limiting the cumulative probability of the selected tokens. A lower value results in more focused responses. A higher value allows for more diverse outputs.
41+
- Max Response Length (Tokens): Sets the maximum number of tokens in the model's response. This limits the length of the generated text.
42+
- Frequency Penalty: Reduces the likelihood of the model repeating the same tokens in the response. Higher values discourage repetition.
43+
- Presence Penalty: Encourages the model to introduce new topics or concepts in the response. Higher values promote diversity in the generated text.
44+
45+
## Chat prompt
46+
47+
![Chat prompt input box in AI Toolkit playground showing an attachment icon for adding files.](./images/playground/chat_prompt.png)
48+
49+
The chat prompt is the input text you provide to the model. You can enter a question, statement, or any text that you want the model to respond to. The model generates a response based on the provided prompt and the configured system prompts and parameters.
50+
51+
Available features in the chat prompt input box:
52+
- **Send**: Submit the chat prompt to the model for processing.
53+
- **Clear**: Clear the chat prompt input box.
54+
- **File Attachment**: Attach files to the chat prompt for multi-modal models.
55+
- **Image Attachment**: Attach images to the chat prompt for multi-modal models.
56+
- **Web Search**: Perform a web search to find information related to the chat prompt.
57+
- **Mode Switch**: Switch between different modes for using AI features.
58+
59+
### Switch between modes
60+
You can switch between different modes in the chat prompt input box:
61+
- **Use native supported AI features from a model**: This mode allows you to use the native capabilities of the selected model, such as web search or file attachment.
62+
- **Use all AI features**: This mode allows you to use AI feature implemented by AI Toolkit, such as file attachment regardless of the model's native capabilities.
63+
64+
If a capability is not supported by the selected model or by AI Toolkit, a warning appears above the chat prompt input box. The warning recommends that you switch to a different model that supports the feature.
65+
66+
![Screenshot showing warning message displayed above the chat prompt input box stating Capability not supported.](./images/playground/capability_warning.png)
67+
68+
## Model responses
69+
![AI Toolkit playground response area showing a generated Python code snippet for encoding text in base64, with options to copy or regenerate the response.](./images/playground/response_area.png)
2370

24-
![Select a model and configure context instructions in the playground.](./images/playground/parameters.png)
71+
Model responses are displayed above the chat prompt input box. Before you enter a chat prompt, the playground shows a welcome screen with three starter prompts for you to get started with the playground.
2572

26-
1. Optionally, add context instructions to guide the model response
73+
Available features in the model response area:
74+
- **Copy Response**: Copy the model's response text to the clipboard.
75+
- **Regenerate Response**: Regenerate the model's response based on the same prompt or regenerate response with another model.
76+
- **Copy Code**: Copy the generated code in the response to the clipboard or insert it into a new file.
77+
- **Rendering**: Playground supports rendering of code snippets, images, markdown, LaTex and other content in the model's response.
78+
- **Token Count**: Displays the number of tokens used in the model's response. This helps you understand the cost and performance of the model.
2779

28-
1. Optionally, configure the model parameters
80+
## Tool bar
2981

30-
1. Enter a chat prompt in the chat input box
82+
The tool bar in the playground provides quick access to various actions and features:
83+
![Tool bar in AI Toolkit playground showing buttons for clearing chat history, saving chat, and switching modes.](./images/playground/tool_bar.png)
3184

32-
From the chat input box, you can also clear chat history or add attachments for the prompt.
85+
- **New Playground**: Create a new playground session.
86+
- **History**: View, rename, delete, and switch to a chat history.
87+
- **Automatic Naming**: Names of chat sessions are automatically generated based on the user prompt.
88+
- **Compare**: Compare responses from different models side by side.
89+
- **View Code**: View the generated code for the chat prompt from LLM integration.
3390

34-
## Add attachments for multi-modal models
91+
### Compare model responses
92+
![Screenshot showing side by side comparison of responses from different models in the AI Toolkit playground.](./images/playground/compare.png)
3593

36-
Multi-modal models are models that can process multiple types of input, such as text, images, audio, or video. By attaching files to your chat prompt, you can ask questions about the contents of these files.
94+
The **Compare** feature allows you to compare responses from different models side by side. This is useful for evaluating the performance of different models on the same prompt.
3795

38-
For the models that support attachments, the attachment icon (paperclip) will show in the chat input box. Select the icon, and follow the instructions to attach one or more local files and use them with your prompt.
96+
To compare model responses:
97+
1. Select the **Compare** button in the tool bar.
98+
2. Choose the models you want to compare from the dropdown list.
99+
3. Enter the same chat prompt in one of the input boxes. Your prompt is copied to the other input boxes.
100+
4. Select **Send** to submit the prompt to all selected models.
101+
5. The responses from each model appear side by side for easy comparison.
39102

40-
![Adding attachments](./images/playground/attachment.png)
103+
For each model you are comparing, you can:
104+
- Copy the response text to the clipboard.
105+
- Configure the model parameters for the selected model.
106+
- Continue the chat with the selected model.

0 commit comments

Comments
 (0)