No such file or directory: data/word2vec

[Smolky]

Hi Daniel. Nice to meet you.

I am tryining to make use of your tool but I have one problem and I hope you could help me pointing me in the right direction.

First of all, I have successlly download the tool, the datasets and run these scripts:

python -m scripts.install_word2vec
python -m scripts.clean_wikitext 

It is worth noting that the first time I downloaded only the Word level documents. But I could not run the script. However, when I also downloaded the raw / character level documents all went fine.

Next, when I run:

run fit myexperiment0 seq2seq wikitext --batch_size=32 --learning_rate=0.001

I get this error

FileNotFoundError: [Errno 2] No such file or directory: 'data/word2vec'

I have checked that a new directory myexperiment0 is created with two files: runpy.json and hparams.json, which content is :

{"model": "seq2seq", "data_loader": "wikitext"}

{
    "attention": "bahdanau",
    "attention_pos": 2,
    "batch_size": 32,
    "beam_width": 5,
    "chunk": false,
    "corpus": 2,
    "dropout": 0.0,
    "embed_size": 300,
    "epochs": 2,
    "hidden_size": 128,
    "learning_rate": 0.0005,
    "max_seq_len": 40,
    "optimizer": "adam",
    "rnn_layers": 2,
    "shuffle": true,
    "vocab_size": 20000
}

I also try with (I deleted the experiment0 folder before)

python run.py fit myexperiment0 seq2seq wikitext_char

but the output was the same.

When the filedata/word2vec is created? It is my responsability to create it or it is automatically created in run?

Kind regards