Bedrock Titan Image Generator G1 content filters

Question

The Service make amazing image but not suitable for production as rejecting most prompt for content filter policy.

`com.amazon.amazonbedrockinferenceruntimeservice.ValidationException: This request has been blocked by our content filters. Our filters automatically flagged this prompt because it may conflict our AUP or AWS Responsible AI Policy. Please adjust your text prompt to submit a new request`

for example a simple prompt of 3 words like : "paris love story" will be refuse

Is there a way to remove the filter policy?

Arjun Arippa · Answer

Hi,

Some of the filters are inbuilt into the LLMs/Bedrock which might not be accessible to the end-customer, while customers can make use of Guardrails to setup custom filters.
Please try changing the prompts, if you're still facing issues. In addition, please feel to try the latest model Image Generator G2.

Thanks,
Arjun

Bedrock Titan Image Generator G1 content filters

新增您的答案

相關內容