# Disable determinism in inference API text generation

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fdisable-determinism-in-inference-api-text-generation%2F13630%3E%5C
**Category:** Beginners\
**Created:** [January 12, 2022, 9:00pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fdisable-determinism-in-inference-api-text-generation%2F13630 "2022-01-12T21:00:48Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![louis030195](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/louis030195/32/3591_2.png) [@louis030195](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flouis030195)%5C
**Post date:** [January 12, 2022, 9:00pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fdisable-determinism-in-inference-api-text-generation%2F13630%2F1 "2022-01-12T21:00:48Z")

</div>

According to the [Inference API documentation](/static-proxy?url=https%3A%2F%2Fapi-inference.huggingface.co%2Fdocs%2Fpython%2Fhtml%2Fdetailed_parameters.html%23text-generation-task)%2C when the option “use\_cache” is set to false, it should not return deterministic output, but does not work for me.

```auto
def huggingface_api_completion(prompt: str) -> str:
    API_URL = f"/static-proxy?url=https%3A%2F%2Fapi-inference.huggingface.co%2Fmodels%2F%7BMODEL_V2%7D"
    headers = {"Authorization": f"Bearer {os.environ.get('HUGGINGFACE_KEY')}"}

    data = json.dumps(
        {
            "inputs": prompt,
            "parameters": {
                "max_length": round(len(prompt) / 5) + 100,
                "num_return_sequences": 1,
                "return_text": False,
                "return_full_text": False,
                "do_sample": True,
                "top_k": 50,
                "top_p": 0.95,
                "end_sequence": "\n"
            },
            "options": {
                "wait_for_model": True,
                "use_cache": False,
            },
        }
    )
    response = requests.request("POST", API_URL, headers=headers, data=data)
    data = json.loads(response.content.decode("utf-8"))
    completions = data[0]["generated_text"].strip()
    return completions

```

How can I enforce non-deterministic outputs? When doing local inference I tweak the seed but here IDK?