Implemented train for testing on my dataset.

[Bill-ai]

corenlp.client.PermanentlyFailedException: Timed out waiting for service to come alive. The model asked for the Type question, when I typed question and the above error pops up? Do you know why this might be a problem? Please help.

PS: Calling the def infer function, Changed the args like --do_train to False --do_infer to True --infer_loop to True --EG to True and in the def infer block changed the args show_table to True and show_anwer_only to True.

Thanks Bill

[guotong1988]

Start a local stanford-corenlp service.

[Bill-ai]

### Thanks, started the server and this pops up. I know there is already an issue created for this, but I didn't understand the answer given by you. Can you please explain it to me in a simpler way?****

Traceback (most recent call last): File "train.py", line 837, in beam_size=1, show_table=False, show_answer_only=True File "train.py", line 667, in infer beam_size=beam_size) File "/home/govind/Documents/Cricket/NL2SQL-BERT/sqlova/model/nl2sql/wikisql_models.py", line 115, in beam_forward knowledge=knowledge, knowledge_header=knowledge_header) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 532, in call result = self.forward(*input, *kwargs) File "/home/govind/Documents/Cricket/NL2SQL-BERT/sqlova/model/nl2sql/wikisql_models.py", line 562, in forward knowledge = [k + (mL_n - len(k)) [0] for k in knowledge] TypeError: 'NoneType' object is not iterable

PS: I'm using the def infer below the input, but you were talking that infer API is not working and thus to use test? Please correct me if I'm wrong.

Best Bill

[guotong1988]

Prepare the data as the readme written.

[Bill-ai]

Took the same example of readMe file and pasted it into ctable.tables.json in data_and_model folder.

python3 train.py 
BERT-type: uncased_L-12_H-768_A-12
Batch_size = 32
BERT parameters:
learning rate: 1e-05
Fine-tune BERT: False
vocab size: 30522
hidden_size: 768
num_hidden_layer: 12
num_attention_heads: 12
hidden_act: gelu
intermediate_size: 3072
hidden_dropout_prob: 0.1
attention_probs_dropout_prob: 0.1
max_position_embeddings: 512
type_vocab_size: 2
initializer_range: 0.02
Load pre-trained parameters.
Seq-to-SQL: the number of final BERT layers to be used: 2
Seq-to-SQL: the size of hidden dimension = 100
Seq-to-SQL: LSTM encoding layer size = 2
Seq-to-SQL: dropout rate = 0.3
Seq-to-SQL: learning rate = 0.001
Type question: what
Traceback (most recent call last):
  File "train.py", line 837, in <module>
    beam_size=1, show_table=False, show_answer_only=True
  File "train.py", line 653, in infer
    hds1 = tb1['header']
KeyError: 'header'

Had a look into train.py the def infer has a variable assigned hsa1 which accesses header. I'm I doing something wrong?

[guotong1988]

You have to debug and recode..

[Bill-ai]

I tried the repo from scratch, but still getting the same error. In ctable.tables.json I have added the same sample data input in the first line. Can you please help me out here? I don't understand if there is no header in your input sample data, but you still want to access it through def infer? How is that possible?

[guotong1988]

https://drive.google.com/file/d/1iJvsf38f16el58H4NPINQ7uzal5-V4v4

the table data is here

[guotong1988]

my sample data is the questions and SQLs. The table data is not uploaded.

[Bill-ai]

Okay, thanks for the clarity, But the thing is I'm actually doing an inference. So I don't need the questions and sqls I guess? I have already downloaded the table data file, I see some tables and jsonl files and uncased vocab and model files also.

Can you tell me what should I insert into ctable.tables.json file to run the main.py file for inference? Or Quick way you can tell is how can I convert the CSV file to get queries directly by running main.py. Please tell me how do I preprocess the CSV file? and what changes do I need to have in main.py? Or send me a file which I can directly run for inference.

I'm sorry for the long conversation. Can I email you up?

[sarkaramal]

Hi, I am also facing the same issue. Is there any update on this. Can u share the test for influence?

[guotong1988]

Inference is not supported now.