Gradio Block() can't detect imported library like numpy in jupyter notebook

[kelvinhammond]

Describe the bug

Exact issue described here is still valid but I cannot reopen this ticket #3625

Gradio fails to pull imports on reload

Have you searched existing issues? 🔎

• I have searched and found no existing issues

Reproduction

import gradio as gr

%%blocks
# Anything in under gr.NO_RELOAD won't be reloaded when the block is re-run (afaik)
if gr.NO_RELOAD:
    import numpy as np
    from transformers import pipeline
    transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base.en")

def transcribe(stream, new_chunk):
    sr, y = new_chunk
    
    # Convert to mono if stereo
    if y.ndim > 1:
        y = y.mean(axis=1)
        
    y = y.astype(np.float32)
    y /= np.max(np.abs(y))

    if stream is not None:
        stream = np.concatenate([stream, y])
    else:
        stream = y

    return stream, transcriber({"sampling_rate": sr, "raw": stream})["text"]

waveform_options = gr.WaveformOptions(
    waveform_color="#01C6FF",
    waveform_progress_color="#0066B4",
    skip_length=2,
    show_controls=False,
)

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            state = gr.State()
            audio = gr.Audio(
                sources=["microphone", "upload"],
                show_download_button=True,
                waveform_options=waveform_options,
                streaming=True,
            )
            with gr.Row():
                clear_btn = gr.ClearButton()
                submit_btn = gr.Button("Submit", variant="primary")
        output = gr.Textbox(label="Output")

    submit_btn.click(
        fn=transcribe,
        inputs=[state, audio],
        outputs=[state, output],
        api_name="transcribe")

    gr.on(
        triggers=[audio.stream],
        fn=transcribe,
        inputs=[state, audio],
        outputs=[state, output],
    )

Screenshot

No response

Logs

Traceback (most recent call last):
  File "/opt/conda/lib/python3.12/site-packages/gradio/queueing.py", line 625, in process_events
    response = await route_utils.call_process_api(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/gradio/route_utils.py", line 322, in call_process_api
    output = await app.get_blocks().process_api(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/gradio/blocks.py", line 2098, in process_api
    result = await self.call_function(
             ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/gradio/blocks.py", line 1645, in call_function
    prediction = await anyio.to_thread.run_sync(  # type: ignore
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/anyio/to_thread.py", line 56, in run_sync
    return await get_async_backend().run_sync_in_worker_thread(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/anyio/_backends/_asyncio.py", line 2461, in run_sync_in_worker_thread
    return await future
           ^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/anyio/_backends/_asyncio.py", line 962, in run
    result = context.run(func, *args)
             ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/conda/lib/python3.12/site-packages/gradio/utils.py", line 883, in wrapper
    response = f(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^
  File "<string>", line 16, in transcribe
NameError: name 'np' is not defined

System Info

Gradio Environment Information:
------------------------------
Operating System: Linux
gradio version: 5.16.0
gradio_client version: 1.7.0

------------------------------------------------
gradio dependencies in your environment:

aiofiles: 23.2.1
anyio: 4.8.0
audioop-lts is not installed.
fastapi: 0.115.8
ffmpy: 0.5.0
gradio-client==1.7.0 is not installed.
httpx: 0.28.1
huggingface-hub: 0.28.1
jinja2: 3.1.5
markupsafe: 2.1.5
numpy: 2.1.3
orjson: 3.10.15
packaging: 24.2
pandas: 2.2.3
pillow: 11.1.0
pydantic: 2.10.6
pydub: 0.25.1
python-multipart: 0.0.20
pyyaml: 6.0.2
ruff: 0.9.6
safehttpx: 0.1.6
semantic-version: 2.10.0
starlette: 0.45.3
tomlkit: 0.13.2
typer: 0.15.1
typing-extensions: 4.12.2
urllib3: 2.3.0
uvicorn: 0.34.0
authlib; extra == 'oauth' is not installed.
itsdangerous; extra == 'oauth' is not installed.


gradio_client dependencies in your environment:

fsspec: 2024.9.0
httpx: 0.28.1
huggingface-hub: 0.28.1
packaging: 24.2
typing-extensions: 4.12.2
websockets: 14.2

Severity

Blocking usage of gradio

[freddyaboulton]

Hi @kelvinhammond - You can place the code inside NO_RELOAD in a separate notebook cell. The NO_RELOAD was designed more for file/script use cases

[kelvinhammond]

Thank you.

It seems the transcribe function must also be in it's own cell but the block is re-executable so this should work for now.

The docs for this should probably mention this as a caveat. https://www.gradio.app/guides/developing-faster-with-reload-mode#jupyter-notebook-magic

[freddyaboulton]

Would you like to open a PR?

[kelvinhammond]

Maybe, not too familiar with gradio yet so I wouldn't be comfortable updating its documentation just yet.