### What happened?

Tried setting output_parse_pii as True with presidio but the response is still masked I tried setting it in the UI, config.yml and even in the completion request as below

``` { "model": "mymodel", "messages": [ {"role": "user", "content": "my name is jane doe"} ], "guardrails": ["pii_mask"], "guardrail_config": {"output_parse_pii":false}

} ```

RESPONSE -- STILL MASKED

``` choices": [ { "finish_reason": "stop", "index": 0, "message": { "content": "Hello, <PERSON>! How can I assist you today?", "role": "assistant", "annotations": [] }, "provider_specific_fields": { "content_filter_results": { "hate": { "filtered": false, ```

### Relevant log output

```shell

```

### Are you a ML Ops Team?

No

### What LiteLLM version are you on ?

1.77.1

### Twitter / LinkedIn details

_No response_