Bert: I'm getting key error '2' when i run run_classifier.py

[sneharaogr]

I got key error 2 when I ran run_classifier.py on my train.tsv file. I'm unable to figure out what could possibly be the reason for this error ? Could someone please help me fix this.

The error that i encountered is exactly :
INFO:tensorflow:_TPUContext: eval_on_tpu True
WARNING:tensorflow:eval_on_tpu ignored because use_tpu is False.
INFO:tensorflow:Writing example 0 of 13
INFO:tensorflow:* Example *
INFO:tensorflow:guid: train-0
INFO:tensorflow:tokens: [CLS] a [SEP]
INFO:tensorflow:input_ids: 101 1037 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
INFO:tensorflow:input_mask: 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
INFO:tensorflow:label: 1 (id = 1)
Traceback (most recent call last):
File "bert/run_classifier.py", line 981, in
tf.app.run()
File "/home/sneha/anaconda3/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 125, in run
_sys.exit(main(argv))
File "bert/run_classifier.py", line 870, in main
train_examples, label_list, FLAGS.max_seq_length, tokenizer, train_file)
File "bert/run_classifier.py", line 490, in file_based_convert_examples_to_features
max_seq_length, tokenizer)
File "bert/run_classifier.py", line 459, in convert_single_example
label_id = label_map[example.label]
KeyError: '2'

Thanks in advance !!

[hsm207]

You are getting this error because the format of your train.tsv is incompatible with the task_name argument you specified for task classifier. Make sure the format of your train.tsv is compatible with the data processor associated with whatever task_name you specified.

[amithadiraju1694]

@hsm207 @Sneha-Rao-GR

I had a similar issue once. I feel that the issue is not necessarily because of the data format.

In my case I had training data in the format of:

<ID> <LABEL> <THROW_AWAY> <SENTENCE>

0 1 a 'Some random sentence'

Yet, I had this issue. After bit of digging, I realized that the run_classifier file is configured by default to work with binary classification, so when the classifier comes across any label beyong 0 or 1 it throws an error for obvious reasons, because of the way it's get_label function is coded.

So, go to -> BERT BASE FOLDER - > run_classifier.py - > get_label function and change it from:

def get_labels(self): """See base class.""" return ["0", "1"]

to

def get_labels(self): """See base class.""" return ["0", "1", "2", "3", "4", "5", "6", "7"] # How many ever distinct labels you have in your data set

I Hope it helps.

[p4rk3r]

@amit8121 - Your comment solved the problem for me. Thanks a million!

[amithadiraju1694]

@amit8121 - Your comment solved the problem for me. Thanks a million!

Glad I could help.

[JainAnvitha]

@amit8121 - how would you solve the same in bert pretrained model?

[ShrikanthSingh]

I am using tfhub module similar to the one in BERT repo example but I have used similar code snippet of train_features for generating test_features also because I have text_b content which is not none in my case. But I am getting a key error 0. I don't know where am i going wrong.
Thanks in advance

test = pd.read_csv('/content/drive/My Drive/Datasets/Test_set US_SUP.csv') test.fillna('No text',inplace=True)

DATA_COLUMN_TEST_A = 'content_A_test' DATA_COLUMN_TEST_B = 'content_B_test' LABEL_COLUMN = 'Labels_test' label_list_test = [x for x in np.unique(test.Labels_test)] print(label_list_test)

test_InputExamples = test.apply(lambda x: bert.run_classifier.InputExample(guid=None, text_a = x[DATA_COLUMN_TEST_A], text_b = x[DATA_COLUMN_TEST_B], label = 0), axis = 1)

test_features = bert.run_classifier.convert_examples_to_features(test_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)

Error:
INFO:tensorflow:Writing example 0 of 15

INFO:tensorflow:Writing example 0 of 15

KeyError Traceback (most recent call last)
in ()
----> 1 test_features = bert.run_classifier.convert_examples_to_features(test_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)

1 frames
/usr/local/lib/python3.6/dist-packages/bert/run_classifier.py in convert_single_example(ex_index, example, label_list, max_seq_length, tokenizer)
457 assert len(segment_ids) == max_seq_length
458
--> 459 label_id = label_map[example.label]
460 if ex_index < 5:
461 tf.logging.info("* Example *")

KeyError: 0

[AbhilashG97]

Hello @ShrikanthSingh, I'm stuck with the same error, were you able to resolve that error?

Thank you

[ShrikanthSingh]

Hello @ShrikanthSingh, I'm stuck with the same error, were you able to resolve that error?

Thank you

Hi @AbhilashG97 I partially remember about resolving this error. Hence I think I figured it out by changing the labels from [1, 2, 3, 4, 5] to [0, 1, 2, 3, 4] because the enumeration for the labels in the run classifier starts from 0 and not from 1. But I am not sure of this solution, because it was long time back.

[AbhilashG97]

Hello @ShrikanthSingh, thank you so much, I was able to fix the error.

I did something quite stupid, I forgot to write the code to map my labels to integer numbers. I mapped my labels to integer numbers starting from 0 and it works fine now. :sparkles:

[Lailiwen]

Hi, I got key error 4 when I ran run_classifier.py on the train.tsv file. Does anyone happen to know how to resolve this?

INFO:tensorflow:Writing example 0 of 145342
I0726 12:31:53.243687 13768 run_classifier.py:487] Writing example 0 of 145342
Traceback (most recent call last):
File "run_classifier.py", line 981, in
tf.app.run()
File "C:\Users\ll\anaconda3\lib\site-packages\tensorflow_core\python\platform\app.py", line 40, in run
_run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
File "C:\Users\ll\anaconda3\lib\site-packages\absl\app.py", line 299, in run
_run_main(main, args)
File "C:\Users\ll\anaconda3\lib\site-packages\absl\app.py", line 250, in _run_main
sys.exit(main(argv))
File "run_classifier.py", line 870, in main
train_examples, label_list, FLAGS.max_seq_length, tokenizer, train_file)
File "run_classifier.py", line 490, in file_based_convert_examples_to_features
max_seq_length, tokenizer)
File "run_classifier.py", line 459, in convert_single_example
label_id = label_map[example.label]
KeyError: '4'

Thanks in advance!

[kanishk-verma-dotcom]

Hello @ShrikanthSingh, thank you so much, I was able to fix the error.

I did something quite stupid, I forgot to write the code to map my labels to integer numbers. I mapped my labels to integer numbers starting from 0 and it works fine now. ✨

@AbhilashG97 I encountered the same issue as @ShrikanthSingh and I even changed the the labels to integer starting from 0

df_v2.category = pd.Categorical(pd.factorize(df_v2.category)[0]) ![image](https://user-images.githubusercontent.com/57866492/98054382-b2afdd00-1e32-11eb-99f4-950a8a361e46.png)

Can you please tell me how do I overcome the same KeyError?