SGLang supports two grammar backends:

• Outlines (default): Supports JSON schema and regular expression constraints.
• XGrammar: Supports JSON schema and EBNF constraints and currently uses the GGML BNF format.

We suggest using XGrammar whenever possible for its better performance. For more details, see XGrammar technical overview.

To use Xgrammar, simply add --grammar-backend xgrammar when launching the server. If no backend is specified, Outlines will be used as the default.

try:
response_content = response.choices[0].message.content
# validate the JSON response by the pydantic model
capital_info = CapitalInfo.model_validate_json(response_content)
print_highlight(f"Validated response: {capital_info.model_dump_json()}")
except ValueError as e:
print(f"Validation error: {e}")

Do not use try except, since we do not want the except to happen and pass CI.

Define Schemas: Directly define a JSON schema or use Pydantic to define a schema.
Validate Responses: Use Pydantic to ensure the response matches expected structure and types.

Give example for both Json directly and Pydantic