Widgets Not Displaying

[Rasputin312]

I am trying to run this attention visualization in my Databricks notebook.   This is my code and this is the error I get:

```

from IPython.display import display, Javascript

import ipywidgets as widgets

from ipywidgets import interact

from transformers import AutoTokenizer, AutoModel, utils

from bertviz import model_view

utils.logging.set_verbosity_error()  # Suppress standard warnings

model_name = "microsoft/xtremedistil-l12-h384-uncased"  # Find popular HuggingFace models here: https://huggingface.co/models

input_text = "The cat sat on the mat"

model = AutoModel.from_pretrained(model_name, output_attentions=True)  # Configure model to return attention values

tokenizer = AutoTokenizer.from_pretrained(model_name)

inputs = tokenizer.encode(input_text, return_tensors='pt')  # Tokenize input text

outputs = model(inputs)  # Run model

attention = outputs[-1]  # Retrieve attention from model outputs

tokens = tokenizer.convert_ids_to_tokens(inputs[0])  # Convert input ids to token string

model_view(attention, tokens)  # Display model view
```

/databricks/python/lib/python3.12/site-packages/transformers/tokenization_utils_base.py:1601: FutureWarning: `clean_up_tokenization_spaces` was not set. It will be set to `True` by default. This behavior will be depracted in transformers v4.45, and will be then set to `False` by default. For more details check this issue: https://github.com/huggingface/transformers/issues/31884 warnings.warn(

  <IPython.core.display.Javascript object>

Does anyone have thoughts on how to actually see the display?  

[koji_kawamura · Accepted solution]

Hi @Rasputin312 !

I was able to render the visualization with bertviz library.

The default moview_view html_action is view that does not work with Databricks notebook. Instead, using the returned HTML, we can visualize the model.

display(model_view(attention, tokens, html_action="return")) 

From model_view doc:

html_action: Specifies the action to be performed with the generated HTML object - 'view' (default): Displays the generated HTML representation as a notebook cell output - 'return' : Returns an HTML object containing the generated view for further processing or custom visualization

The entire notebook code is available here. https://gist.github.com/koji-kawamura-db/49c52d5d96b25b936e038fc21d3bbcec