sean-blank / amrcoref Goto Github PK

This above error showed up when I attempted to run 'prepare_data.py'
I checked the version of penman, but its the same version as 'pipreqs' says was used in the codebase. (I ran pipreqs to check for version used in your code; and it is the same as the version I have)
Please help

When running python train.py, it says that:
FileNotFoundError: [Errno 2] No such file or directory: './data/corpora_base/evl'

Dataloader needs to load data in JSON format, but not very clear on how to generate/preprocess from the raw LDC2020T02 data. What the JSON data should look like? And what should be done to get the code running given the LDC2020T02 data txt format?

I used penman 0.6.0 do run prepare_data.py

Then when I ran preprocess.py, I got:

Traceback (most recent call last): File "preprocess.py", line 354, in <module> item = pre_to_json(data_per_doc, links_per_doc, file_name) File "preprocess.py", line 307, in pre_to_json data, cluster = get_clusters_info(links, data) File "preprocess.py", line 260, in get_clusters_info pdb.set_trace() File "preprocess.py", line 240, in mapping_edges assert len(amr.nodes) == len(concept) == len(amr.node_values) AssertionError

I am using the preprocessed data to train and facing the following issue. I understand that it is issue with not finding the model. Any way this can be solved?

PU available: True CuDNN: True
Using GPU To Train... GPU ID: 1
Log file path: ./ckpt/coref.amr.log
Model name './data/bert-base-cased' was not found in model name list (bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese). We assumed './data/bert-base-cased' was a path or url but couldn't find any file associated to this path or url.
load train data
Traceback (most recent call last):
File "train.py", line 165, in
train(args)
File "train.py", line 34, in train
train_data, dev_data, test_data, vocabs = make_data(args, tokenizer)
File "/home/researcher/Trina/SeanBlankAMRcoref-main/dataloader.py", line 504, in make_data
train_data = load_json(args.train_data, args, tokenizer)
File "/home/researcher/Trina/SeanBlankAMRcoref-main/dataloader.py", line 156, in load_json
token_bert_ids = get_bert_ids(tokens, args, tokenizer)
File "/home/researcher/Trina/SeanBlankAMRcoref-main/dataloader.py", line 93, in get_bert_ids
for char in tokenizer.tokenize(word):
AttributeError: 'NoneType' object has no attribute 'tokenize'

Hello,

I ran the training and eval script for one time and got F1 at around 54, which is much lower than the reported number in the paper, 62. Is that number obtained by choosing the best across multiple runs? If yes, how many times do you have to run?

Hi, I'm trying to prepare data (we bought an AMR3.0 licence).
But I failed to run preprocess.py. It seems that data/align_unsplit does not exist. Which files must I copy there? In your data/ there are three msamr_dfb_*.align, but I do not have these files in my AMR3.0 data. Are these renamed files or produced by a script on the AMR3 data?

I find that the code in prepare_data.py , did nothing with align_path , so no file generate under align_unsplit.

可以发处理好的数据给我吗，我也遇到了和之前提问者相同的问题，我学校有LDC的license，可以的话请发到[email protected]，谢谢

Hello, when I try to use Bert, error comes out that mat1 and mat2 shapes cannot be multiplied, so I want to ask how you run the code using Bert?

Please provide the exact versions for the packages like transformers, etc. Only python and torch have been provided.