Welcome to forums.fast.ai

Part 1 (2017)
274

Hi, I hope I’m posting in the right channel (if not please let me know: it’s for an open science/open data cause!).

We are happy to host the COVEO DATA CHALLENGE @ SIGIR 2021 - you can test your ML skills on a huge, new, rich eCommerce dataset that we just released to the AI community, and you’ll get a chance to write a research paper and present it at the conference!

DATES

  • Registration ends: June 1st, 2021
  • Final leaderboard: June 15th, 2021
  • SIGIR eCom Full day Workshop: July 15th, 2021

If you want to know more, ping me anytime - we also have a Slack channel you can join!

275

How do I post in this forum?

276

Hey all!

Second stab at completing the course - ran into some troubles with the cloud setup the first time and gave up almost immediately. Taking Jeremy’s suggestion to write a blog this time around and it definitely forces you to think about things!

278

Hi all, thank you, folks, for this course! I am glad to share with you my works.

279

Running this

from fastbook import *
urls = search_images_ddg('grizzly bear', max_images=100)
len(urls),urls[0]

and got this error

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-4-a727f76d86a1> in <module>
      1 from fastbook import *
----> 2 urls = search_images_ddg('grizzly bear', max_images=100)
      3 print(len(urls))

/opt/conda/lib/python3.7/site-packages/fastbook/__init__.py in search_images_ddg(term, max_images)
     55     assert max_images<1000
     56     url = 'https://duckduckgo.com/'
---> 57     res = urlread(url,data={'q':term}).decode()
     58     searchObj = re.search(r'vqd=([\d-]+)\&', res)
     59     assert searchObj

AttributeError: 'str' object has no attribute 'decode'

Searched it and what I understood was that its already decoded . What to do next?

280

saw the source code at https://github.com/fastai/course20/blob/master/fastbook/init.py and its correct there

def search_images_ddg(term, max_images=200):
    "Search for `term` with DuckDuckGo and return a unique urls of about `max_images` images"
    assert max_images<1000
    url = 'https://duckduckgo.com/'
    res = urlread(url,data={'q':term})
    searchObj = re.search(r'vqd=([\d-]+)\&', res)
    assert searchObj
    requestUrl = url + 'i.js'
    params = dict(l='us-en', o='json', q=term, vqd=searchObj.group(1), f=',,,', p='1', v7exp='a')
    urls,data = set(),{'next':1}
    while len(urls)<max_images and 'next' in data:
        try:
            data = urljson(requestUrl,data=params)
            urls.update(L(data['results']).itemgot('image'))
            requestUrl = url + data['next']
        except (URLError,HTTPError): pass
        time.sleep(0.2)
    return L(urls)

So I think I should update my fastbook module in the notebook.

281

checked the version, its already the latest (0.0.16)

282

Recently I came across a fatraven site where I came across this AI-powered Recruitment Lead Generation Tool so how AI can help recruitment?

283

Yeah just replying for getting posting privilidges too

284

How long does it take to be able to post questions?

285

I just got posting privileges right after posting the above

Here’s what I did:
-Waited around 1 day
-Liked a few comments
-Made several replies

Hopeufully this helps

1 Like
286

I have started fastai book and was going through the first notebook 01_intro.ipynb. The imdb text classification was taking a long time to run.

from fastai.text.all import *

dls = TextDataLoaders.from_folder(untar_data(URLs.IMDB), valid='test', bs=16)
learn = text_classifier_learner(dls, AWD_LSTM, drop_mult=0.5, metrics=accuracy)
learn.fine_tune(4, 1e-2)

Is this common?

287

I am actually getting an error @jeremy Any help,

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-1-c1b52624de93> in <module>()
      3 dls = TextDataLoaders.from_folder(untar_data(URLs.IMDB), valid='test', bs=16)
      4 learn = text_classifier_learner(dls, AWD_LSTM, drop_mult=0.5, metrics=accuracy)
----> 5 learn.fine_tune(4, 1e-2)

18 frames
/usr/local/lib/python3.7/dist-packages/torch/autograd/__init__.py in backward(tensors, grad_tensors, retain_graph, create_graph, grad_variables, inputs)
    147     Variable._execution_engine.run_backward(
    148         tensors, grad_tensors_, retain_graph, create_graph, inputs,
--> 149         allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag
    150 
    151 

RuntimeError: transform: failed to synchronize: cudaErrorIllegalAddress: an illegal memory access was encountered
288

Ahh, that’s why I can’t find a create post button anywhere I guess?

289

You can see in the book that it took them ~3 minutes per epoch. So about 15 minutes total. On my not as fast GPU it takes longer. If you spend much longer than that, you should check that the GPU is actually used (so not CPU)

290

@Matsemann Thanks so much for your reply. If I run this on a colab notebook with a GPU enabled it takes around 1 hour to run. What is the GPU config you are using ? Also if we train a keras model using IMDB it takes only a couple of minutes.

291

how to fix this error?
thanks!

image
image2024×838 121 KB

292

Hi Jessica
AI recruitment is nothing new. It was used in the 1980’s. Anybody can be a salesman so they believe and hundreds apply for the job. We had a system run by a US University for which all the candidates completed a questionnaire and the computer selected the most suitable candidates for human interviews.
Regards Conwyn

293

Hi LisaLi

Unfortunately the course will require many different skills. So each user will have different needs. For somebody with a day job I would say about 12 months.

Regards Conwyn

294

Hey this is a reply so I can ask a question in a new post.