KeyArgo
/
smolagents
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add system prompt modification in building_good_agents

This commit is contained in:
Aymeric 2025-01-06 14:18:56 +01:00
parent 4cfe4a6573
commit 57061ba926
1 changed files with 18 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
docs/source/en/tutorials/building_good_agents.md
View File

@ -168,7 +168,7 @@ Final answer:
/var/folders/6m/9b1tts6d5w960j80wbw9tx3m0000gn/T/tmpx09qfsdd/652f0007-3ee9-44e2-94ac-90dae6bb89a4.png /var/folders/6m/9b1tts6d5w960j80wbw9tx3m0000gn/T/tmpx09qfsdd/652f0007-3ee9-44e2-94ac-90dae6bb89a4.png
``` ```
The user sees, instead of an image being returned, a path being returned to them. The user sees, instead of an image being returned, a path being returned to them.
It could look like a bug from the system, but actually the agentic system didn't cause the error: it's just that the LLM engine did the mistake of not saving the image output into a variable. It could look like a bug from the system, but actually the agentic system didn't cause the error: it's just that the LLM brain did the mistake of not saving the image output into a variable.
Thus it cannot access the image again except by leveraging the path that was logged while saving the image, so it returns the path instead of an image. Thus it cannot access the image again except by leveraging the path that was logged while saving the image, so it returns the path instead of an image.
The first step to debugging your agent is thus "Use a more powerful LLM". Alternatives like `Qwen2/5-72B-Instruct` wouldn't have made that mistake. The first step to debugging your agent is thus "Use a more powerful LLM". Alternatives like `Qwen2/5-72B-Instruct` wouldn't have made that mistake.
@ -191,6 +191,22 @@ If after trying the above, you still want to change the system prompt, your new
- `"{{managed_agents_description}}"` to insert the description for managed agents if there are any. - `"{{managed_agents_description}}"` to insert the description for managed agents if there are any.
- For `CodeAgent` only: `"{{authorized_imports}}"` to insert the list of authorized imports. - For `CodeAgent` only: `"{{authorized_imports}}"` to insert the list of authorized imports.
Then you can change the system prompt as follows:
```py
from smolagents.prompts import CODE_SYSTEM_PROMPT, HfApiModel
modified_system_prompt = CODE_SYSTEM_PROMPT + "\nHere you go!" # Change the system prompt here
agent = CodeAgent(
tools=[],
model=HfApiModel(),
system_prompt=modified_system_prompt
)
```
This also works with the ToolCallingAgent.
### 3. Extra planning ### 3. Extra planning
@ -203,7 +219,7 @@ from dotenv import load_dotenv
load_dotenv() load_dotenv()
# Import tool from Hub # Import tool from Hub
image_generation_tool = load_tool("m-ric/text-to-image", cache=False) image_generation_tool = load_tool("m-ric/text-to-image", trust_remote_code=True)
search_tool = DuckDuckGoSearchTool() search_tool = DuckDuckGoSearchTool()