problems in "Search Similar Images" doc

[mgogoulos]

Hi, having installed jina through pip (v 2.6.4), I tried to follow https://docs.jina.ai/datatype/image/image2image/ but it fails

I guess this is deprecated?

from jina.types.document.generators import from_files
...
ModuleNotFoundError: No module named 'jina.types.document'

Nex, FlashImageEmbedder is not defined/imported before

.add(uses=FlashImageEmbedder, name="Encoder")
...
NameError: name 'FlashImageEmbedder' is not defined

Thanks

[JoanFM]

Hello @mgogoulos ,

Sorry for the inconvenience.

For the first problem:

• This is our fault for not clarifying on what versions were the examples working. I suggest you go to some previous versions 2.5.
• We will soon work on updating these tutorials since we are planning a new release soon.

For the second problem:

• It seems there is a missmatch, FlashImageEmbedder should be FlashImageEncoder which is the executor that was actually defined.

Sorry for the inconvenience

[mgogoulos]

Hi, kudos for the lightning-fast reply!

I now install all dependencies for this doc using

pip install jina==2.5 ipython vissl fairscale lightning-flash[image]

perform the small change (FlashImageEmbedder to FlashImageEncoder)

but now end up on a dependency chaos

        Encoder@69249[C]:can not load the executor from FlashImageEncoder
        Encoder@69249[E]:ExecutorFailToLoad() during <class 'jina.peapods.runtimes.zmq.zed.ZEDRuntime'> initialization
    self._embedder = ImageEmbedder(embedding_dim=1024)
TypeError: __init__() got an unexpected keyword argument 'embedding_dim'

The pip install command brought me

lightning-flash             0.6.0
pytorch-lightning           1.5.10

which is clearly not what it needs.

Suggestions on what versions of flash to install? Even better would be to get the exact environment this needs to work!

Thanks

[JoanFM]

Hey @mgogoulos

Sorry for the late reply, we have been working on releaing 3.0 and putting all these tutorials into the Learning Portal and should be soon available.

@alexcg1 @CatStark can you please point to the proper section?

[alexcg1]

@Shubhamsaboo is leading these efforts. He can answer better than me

[Shubhamsaboo]

Hey, @mgogoulos!

Sorry for the trouble with the tutorial, we were in the process of migrating them from Jina 2 to Jina 3!

You can find the working colab notebook for the same tutorial -> https://colab.research.google.com/github/jina-ai/tutorial-notebooks/blob/main/image2image.ipynb

Soon these changes with all other tutorials will be reflected on our Learning Bootcamp.

Hope this helps and thanks for being patient!

[mgogoulos]

I will give it a try, thanks a lot for replying here

[CatStark]

Hi @mgogoulos ,

No problem. I'll close the issue now but feel free to open a new one if we can help with anything else :)