Hi Matt that is correct if.
Once you get it working on your desktop the fun begins when you deploy it.
The render guide is good however it is likely you may have a few issues deploying it. So check the posts on this thread.
On point 3 your library versions on your laptop may be different to the repository. So you may have to change the requirements.txt when you deploy the code on render.com. See the threads which contain !pip list for an explanation.
Cheers mrfabulous1:smiley:
If want to host you model on Google Drive and still have the problem of not being able to download the large, this is the code I´ve used on server.py, it replaces the function download_file():
Note that now instead of the url, you just pass the id of the file:
model_file_id = '19xqtsyusFddcSkdCm1hlhW3y6ljGck5L'
def download_file_from_google_drive(id, destination):
if destination.exists():
return
URL = "https://docs.google.com/uc?export=download"
session = requests.Session()
response = session.get(URL, params = { 'id' : id }, stream = True)
token = get_confirm_token(response)
if token:
params = { 'id' : id, 'confirm' : token }
response = session.get(URL, params = params, stream = True)
save_response_content(response, destination)
def get_confirm_token(response):
for key, value in response.cookies.items():
if key.startswith('download_warning'):
return value
return None
def save_response_content(response, destination):
CHUNK_SIZE = 32768
with open(destination, "wb") as f:
for chunk in response.iter_content(CHUNK_SIZE):
if chunk: # filter out keep-alive new chunks
f.write(chunk)
async def setup_learner():
download_file_from_google_drive(model_file_id, path/'models'/f'{model_file_name}.pth')
data_bunch = ImageDataBunch.single_from_classes(path, classes,
ds_tfms=get_transforms(), size=224).normalize(imagenet_stats)
learn = cnn_learner(data_bunch, models.resnet34, pretrained=False)
learn.load(model_file_name)
return learn
Did Render start charging fastai students? I noticed an email telling me to add a credit card and also an unpaid bill.
Yes, Render has always charged for compute, but a few months ago we added $25 in credits to new accounts. I believe you got an email because your credits ran out.
Thanks for clarifying.
Hello everyone! I’ve deployed my model on Render but the website seems to be stuck on the “Analyzing…” mode. Looking through the logs, I do see prediction = learn.predict(img)[0] so I know it’s predicting, but no result shows on the website.
The error in the Render logs I get is AttributeError: ‘Conv2d’ object has no attribute ‘padding_mode’. Any help / advice?
Thanks in advance and thanks @anurag for Render!
I get this same error when trying to predict:
img = data.test_ds[0][0]
learn.predict(img)
~/.conda/envs/fastai_v1_2/lib/python3.7/site-packages/fastai/torch_core.py in <listcomp>(.0)
325 def grab_idx(x,i,batch_first:bool=True):
326 "Grab the `i`-th batch in `x`, `batch_first` stating the batch dimension."
--> 327 if batch_first: return ([o[i].cpu() for o in x] if is_listy(x) else x[i].cpu())
328 else: return ([o[:,i].cpu() for o in x] if is_listy(x) else x[:,i].cpu())
329
AttributeError: 'list' object has no attribute 'cpu'
Did you ever work out what was wrong?
Hi wigglepuff hope all is well!
Having created a number of different models now, all but one of the faults when deploying the model assuming it works in development, have been resolved in one of two ways, unless I made a mistake in configuring the model or building the app.
The solution to this if you are using the teddy bear template as the basis for your app, is to use !pip list and confirm that the library versions used in your development environment that you trained the model in are the same on render. You normally have to change the versions in the requirements.txt. search this forum topic for !pip list. for further details.
My solution test in all the top browsers.
Hope this helps mrfabulous1
I had expected it was a version discrepancy but had no idea how to address it, thank you @mrfabulous1!! Did exactly as you recommended and it worked perfectly.
Now I can finally help my friends decide if they look more like Brad Pitt or Elon Musk 
(Thanks again! Wow!)
You’re Welcome.
mrfabulous1
Hi goldenflatulence hope all is well!
1. Have you tried a different browser?
2. What errors do you have in the render console?
3. On which platform did you train your model?
With the above information the problem should be easier to solve.
Cheers mrfabulous1
Hi goldenflatulence hope you are well.
As you say you haven’t noticed any errors, it makes things a little bit more difficult to find the fault.
Please check the following.
can you check the versions of libraries used on GCP and the ones used on render? See posts on using the command “!pip list” in this. thread to do this. If any are different change your library version numbers in requirements.txt file in the render app.
Can you confirm that your model file (.pkl) is being accessed correctly?
Can you check the network in the developer tools of the browser you may find some information when you try to analyze an image?
How large is your model (.pkl) file?
Does the same thing happen with very small images?
Hope these suggestions help!
Cheers mrfabulous1
So, I followed you directions and also read a lot of this thread and its finally working.
Thank you so much.
Hi, guys!
I have next problem: when deploy teddybears - everything is fine. As soon as I change export.pkl file and labels - get net error in log console
Hi volcanoflash hope you are well!
- I have next problem: when deploy teddybears - everything is fine.
Do you mean you have deployed the teddy repository and it works fine on Render?
- As soon as I change export.pkl file and labels
Do you mean that you have trained a new model and created a export.pkl file on a different platform like your desktop or a service provider like Google Colab?
and you have put your export.pkl on google shared drive.
- get net error in log console
You get get the error when you have done 1. and 2. and try to deploy the app.
If you have done exactly what I describe in steps 1,2 and 3 a good start would be to see posts in this thread which contain ‘pip list’ and do as they say.
Cheers mrfabulous1
if your answer two both tho
Does anyone know what the URI is for torchvision 0.2.1? It’s the version I used to export my pkl on Paperspace and I can’t seem to find it on the Pytorch website. I’m having the same issues with the teddy bear classifier repo and version numbers not matching up
Hi korlandril hope all is well!
I haven’t found a specific URL for torchvision 0.2.1
However here are some things you can try,
pip install torchvision==0.2.1
or you can add the following line
RUN pip install torchvision==0.2.1
After the requirements line in the file called “Dockerfile” located in the repository.
One app uses the following requirements.txt however sometimes minor changes are made in one library, which affects another library and its not always documented, so I have quite a few versions of requirements.txt.
numpy==1.16.1
torchvision==0.2.1
https://download.pytorch.org/whl/cpu/torch-1.0.1.post2-cp37-cp37m-linux_x86_64.whl
fastai==1.0.54
starlette==0.12.0
uvicorn==0.4.6
python-multipart==0.0.5
aiofiles==0.4.0
aiohttp==3.5.4
Hope this helps mrfabulous1 `
