claritycafe / ivy Goto Github PK

We'll probably need collaboration with a Monika-centric DDLC Mod as we require a solid database for Monika.

Here's some data points to gather for the corpus :

• @lilmonix3 Twitter Data
• Mainline DDLC
• Monika After Story
• Related Citations

Backtrace :

  File "/projects/MonikaDesu-Rebecca/src/data_utils.py", line 137, in prepare_custom_data
    data_to_token_ids(train_enc, enc_train_ids_path, enc_vocab_path, tokenizer)
  File "/projects/MonikaDesu-Rebecca/src/data_utils.py", line 112, in data_to_token_ids
    vocab, _ = initialize_vocabulary(vocabulary_path)
  File "/projects/MonikaDesu-Rebecca/src/data_utils.py", line 87, in initialize_vocabulary
    rev_vocab.extend(f.readlines())
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/lib/io/file_io.py", line 131, in readlines
    s = self.readline()
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/lib/io/file_io.py", line 124, in readline
    return compat.as_str_any(self._read_buf.ReadLineAsString())
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/util/compat.py", line 106, in as_str_any
    return as_str(value)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/util/compat.py", line 84, in as_text
    return bytes_or_text.decode(encoding)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x92 in position 1: invalid start byte

Looks like the new Model is written under Python 2.7 judging from the syntax, probs need to be rewritten again

@Ovyerus, would you like to take over?

Following the instructions outlined within the 'Setting up Ivy' wiki page, downloading the dependencies and placing Fern's data within the resources folder, running setup.py results in a traceback error:

Traceback (most recent call last): File "F:\Ivy\setup.py", line 1, in <module> from ChatLearner.chatbot.bottrainer import BotTrainer ModuleNotFoundError: No module named 'ChatLearner.chatbot'

Seeing this, I decided to dig within the ChatLearner folder, and nothing was there. Do I have to download files that weren't outlined within the instructions elsewhere?

• Convert to the lukalabs fork we made.
• Monika Corpus (MAS, Vanilla, lilmonix3 twitter data, sanitized related citations)
• Emotional analysis Engine implementation

• Monika After Story Client Injector (Seperate Repository - Link)
• CakeChat + Deep Emotional Analysis Engine
• Monika Corpus Extended.

This was in the original concept, though won't be part of 1.0 GA.

• Tacotron 1.0 Implementation
• Voice Model (Probably contact Jilian Ashcraft for this)

First we need to consider what kind of NN Model we should do for rebecca.

I'm considering a Recurrent model as it allows to listen to the context vector while learning from user input.

Based on what @Oxylibrium relayed on Discord:

• Seems the current tokenizer of CakeChat can't take the shrug ASCII art.
• It's read as 9 tokens.

This is obviously bad but @Martmists suggested a solution for it by having delimiters. This would serve as their issue update log