[Example Request]

I was wondering how to deal with the error below.

It occured when the inference bigins because autopilot created random models.

Received client error (406) from model with message "Accept type 'application/json' is not supported."

To deal with the problem, i fixed the code something like below but is it a right approach?

```
import boto3
from botocore.exceptions import ClientError

runtime = boto3.client("sagemaker-runtime")

def invoke_with_fallback(endpoint_name, payload):
try:
response = runtime.invoke_endpoint(
EndpointName=endpoint_name,
ContentType="application/json",
Accept="application/json",
Body=payload
)
return response["Body"].read().decode("utf-8")

except ClientError as e:
error_code = e.response["Error"]["Code"]
error_msg = e.response["Error"]["Message"]

if error_code == "ModelError" and "application/json" in error_msg:
print("[WARN] JSON に非対応のため CSV で再試行します")

response = runtime.invoke_endpoint(
EndpointName=endpoint_name,
ContentType="text/csv",
Accept="text/csv",
Body=convert_json_to_csv(payload)
)
return response["Body"].read().decode("utf-8")
else:
raise e

def convert_json_to_csv(payload):
import json
data = json.loads(payload)
if isinstance(data, dict):
return ",".join(str(v) for v in data.values())
elif isinstance(data, list):
return "\n".join(",".join(str(v) for v in row.values()) for row in data)
else:
raise ValueError("JSON payload format not supported.")

json_payload = '{"year":2023,"month_sin":0.5,"month_cos":0.86, ... }'
result = invoke_with_fallback("your-endpoint-name", json_payload)
print(result)
```

The issue names no repository files, notebooks, or tests. Start by reviewing the SageMaker runtime invocation described in the message and identifying which example or documentation should address content-type compatibility. Done would require a maintainer-confirmed scope and guidance for handling endpoints with different input and output formats.

Written by the indexing model from the issue text.