You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The issue you're experiencing with the deepseek-chat tool and the 'current_time' function seems to be related to how the tool is responding with multiple instances of "已使用 current_time" before providing the correct date or time. This behavior might be due to how the tool is integrated or how the deepseek model is handling the tool calls.
Here are a few things you might want to check or try:
Tool Configuration: Ensure that the 'current_time' tool is correctly configured with the appropriate parameters, such as the desired time format and timezone. Misconfiguration might lead to unexpected behavior.
Model Configuration: Verify the configuration settings for the deepseek model, such as temperature, top_p, and max_tokens, as these can affect how the model generates responses [1].
Error Handling: Check if there are any error handling mechanisms in place that might be causing the tool to retry or respond multiple times. Dify provides features like retry on failure and error handling strategies that could be influencing the tool's behavior [2][3].
Logs and Debugging: Utilize the logging capabilities of the Dify platform to observe the interactions and identify any anomalies in the tool's execution. This can help pinpoint where the issue might be occurring [4].
If these steps do not resolve the issue, it might be helpful to review the integration of the 'current_time' tool within the codebase to ensure there are no underlying bugs or misconfigurations.
Self Checks
Dify version
0.14.2
Cloud or Self Hosted
Self Hosted (Source)
Steps to reproduce
✔️ Expected Behavior
❌ Actual Behavior
Sometimes there is no answer from LLM after tool call, and sometimes there is a correct answer after 5 iterations tool calls
The text was updated successfully, but these errors were encountered: