Rossmann Data

Part 1 (2018)
1

Can we use differential learning rates for training neural networks on structured data like rossmann.

2

Hi Umar

the Rossman data is not to be found at files.fast.ai, do you know where it is. Thanks for your assistance

3

It is here http://files.fast.ai/part2/lesson14/rossmann.tgz

1 Like
4

Yes, you can. StructuredModel has 3 groups of layers (embeddings, linear + batch norm and output).

5

thanks

6

In lesson 6 rossmann, i"m not able to run the “roassman_data_clean” in google colab. I face this error(below) when i tried to run this code:

PATH=Config().data_path()/Path(‘rossmann/’)
! wget -P {PATH} http://files.fast.ai/part2/lesson14/rossmann.tgz
table_names = [‘train’, ‘store’, ‘store_states’, ‘state_names’, ‘googletrend’, ‘weather’, ‘test’]
tables = [pd.read_csv(PATH/f’{fname}.csv’, low_memory=False) for fname in table_names]
train, store, store_states, state_names, googletrend, weather, test = tables
len(train),len(test)

the error:

hrehbr4
hrehbr4.PNG1051×626 41.8 KB

@bny6613

7

Hi.
You didn’t unzip the rossmann.tgz file, so there is no file named train.csv

You need to use the command below after ! wget .... to unzip
!tar -xzf rossmann.tgz -C ./data/rossmann/

2 Likes
8

oh i’ll try it.
thanks