# How to input negative prompt with Flax Stable Diffusion Img2Img Pipeline when using diffusers?

**URL:** <https://discuss.huggingface.co/t/how-to-input-negative-prompt-with-flax-stable-diffusion-img2img-pipeline-when-using-diffusers/43379>\
**Category:** 🧨 Diffusers\
**Created:** [June 15, 2023, 12:34pm UTC](https://discuss.huggingface.co/t/how-to-input-negative-prompt-with-flax-stable-diffusion-img2img-pipeline-when-using-diffusers/43379 "2023-06-15T12:34:52Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![aeroxy](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/aeroxy/32/16867_2.png) [@aeroxy](https://discuss.huggingface.co/u/aeroxy)\
**Post date:** [June 15, 2023, 12:34pm UTC](https://discuss.huggingface.co/t/how-to-input-negative-prompt-with-flax-stable-diffusion-img2img-pipeline-when-using-diffusers/43379/1 "2023-06-15T12:34:52Z")

</div>

```python
prompt = "masterpiece, best quality, 1girl, at dusk"

neg_prompt = "(low quality, worst quality:1.4), (bad anatomy), (inaccurate limb:1.2), bad composition, inaccurate eyes, extra digit, fewer digits, (extra arms:1.2)"

num_samples = jax.device_count()

pipeline, params = FlaxStableDiffusionImg2ImgPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4",
    revision="flax",
    dtype=jnp.bfloat16,
    safety_checker=None,
    load_attn_procs="sayakpaul/civitai-light-shadow-lora",
)

rng = create_key(0)
rng = jax.random.split(rng, num_samples)
prompt_ids, processed_image = pipeline.prepare_inputs(
    prompt=[prompt] * num_samples,
    image=[init_img] * num_samples,
)

neg_prompt_ids, processed_image = pipeline.prepare_inputs(
    prompt=[neg_prompt] * num_samples,
    image=[init_img] * num_samples,
)

p_params = replicate(params)
prompt_ids = shard(prompt_ids)
processed_image = shard(processed_image)

output = pipeline(
    prompt_ids=prompt_ids,
    image=processed_image,
    neg_prompt_ids=neg_prompt_ids,
    params=p_params,
    prng_seed=rng,
    strength=0.6,
    num_inference_steps=50,
    jit=True,
    height=768,
    width=512,
).images

```

I’ve tried generating neg\_prompt\_ids like prompt\_ids because prepare\_inputs only takes two args. But the above code will crash (without any error).