The Safetensor files that you provided are not valid. Make sure the model weights are valid and that you can load them with the Safetensor method

I apologize for the difficulty you're experiencing with importing the DeepSeek model into Amazon Bedrock. The error you're encountering suggests there might be an issue with the Safetensor files you're trying to use. Here are a few steps you can take to troubleshoot and potentially resolve this issue:

1. Double-check that you're using the correct model files from the DeepSeek repository on Hugging Face. Make sure you've downloaded all the necessary files for the specific DeepSeek model you're trying to import.

2. Verify that the files haven't been corrupted during download or transfer to your S3 bucket. You might want to try re-downloading the model files and re-uploading them to your S3 bucket.

3. Ensure that you're following the correct process for preparing the model package. This typically involves downloading the required files from Hugging Face and uploading them to an S3 bucket in a specific structure.

4. When importing the model into Amazon Bedrock, make sure you're providing the correct S3 path to your model artifacts. The path should be in the form s3://<your-bucket>/folder-with-model-artifacts/.

5. If you're still encountering issues, you might want to try with a different version of the DeepSeek model. DeepSeek offers various distilled versions of their models, ranging from 1.5 to 70 billion parameters. A different version might be more compatible with the import process.

6. As a last resort, you could consider using Amazon SageMaker to fine-tune the model first, and then import it into Amazon Bedrock. This approach might help bypass some compatibility issues.

If none of these steps resolve the issue, I recommend reaching out to AWS support for more specific assistance, as they would have more detailed insight into the Custom Model Import process and any known issues with particular models.

Remember, when working with custom model imports, it's important to ensure you're using supported model formats and following best practices for model versioning and deployment in Amazon Bedrock.
Sources
Deploy DeepSeek-R1 distilled Llama models with Amazon Bedrock Custom Model Import | AWS Machine Learning Blog
Deploy DeepSeek-R1 Distilled Llama models in Amazon Bedrock | AWS Machine Learning Blog
Import a question answering fine-tuned model into Amazon Bedrock as a custom model | AWS Machine Learning Blog