₦airaland Forum

BenedictAbajue:
What version did you use so I know what I'm downgrading to exactly?
mehhhn! Rasa must be pretty old

Pls what IDE did you use ?
VSCode, Spyder or the one that comes with Python

3.9...

Vscode

If it gets too complex just use google collab..

Kobranet:
3.9...

Vscode

If it gets too complex just use google collab..

Okayyyy finally got rasa to work.
Its pretty dumb tho. I need training data now.

Any tips on how to do this plssss?😉😉😉

BenedictAbajue:
Okayyyy finally got rasa to work.
Its pretty dumb tho. I need training data now.

Any tips on how to do this plssss?😉😉😉

Finally i love this!!!

It is dumb because no flow has been added to it...

Scrap the data u want i.e nigeria or any other..then check hugging face for data sets on conversations add ur scraped data to it so that ur model can have a sufficient data to train on...u need data skills ooo data processing is important ooo...

Kobranet:
Finally i love this!!!

It is dumb because no flow has been added to it...

Scrap the data u want i.e nigeria or any other..then check hugging face for data sets on conversations add ur scraped data to it so that ur model can have a sufficient data to train on...u need data skills ooo data processing is important ooo...

yesss

I have basic skills in data processing sha but I never really entered csv, json or spreadsheet stuffs sha

Pls how much data is sufficient?

I'm still new to web-scraping but I've used selenium and beautifulsoup before.

BenedictAbajue:
yesss

I have basic skills in data processing sha but I never really entered csv, json or spreadsheet stuffs sha

Pls how much data is sufficient?

I'm still new to web-scraping but I've used selenium and beautifulsoup before.

What's the purpose of ur chatbot?if it's conversational like chatgpt then u need a Corpus of text starting from about half a million and above.

Kobranet:
What's the purpose of ur chatbot?if it's conversational like chatgpt then u need a Corpus of text starting from about half a million and above.

wowwww

half a million question and answers ...jeeeez!

What if I limit it to specific topics, it would reduce the amount of storage but I have to spend time sorting the datasets

BenedictAbajue:
wowwww

half a million question and answers ...jeeeez!

What if I limit it to specific topics, it would reduce the amount of storage but I have to spend time sorting the datasets

That's the best narrow it down to a specific domain. That way u will use a smaller data set.
Again browse hugging face websites for data.

Kobranet:
That's the best narrow it down to a specific domain. That way u will use a smaller data set.
Again browse hugging face websites for data.

I went there but I did not know what exactly to download. I got one example data from rasa website. I had to wait almost an hour to train and load 25kb dataset.
I was shocked.

Kobranet:
That's the best narrow it down to a specific domain. That way u will use a smaller data set.
Again browse hugging face websites for data.

Is it from their github page?

I was able to get some basic ones from chatgpt sha
Like I asked it for 50 example responses for the thanks intent. And so on

I still need to learn how to use entities, slots and actions.

Then I'm also having problems on how to write the stories since I don't know what users may ask after each turn in the conversation.

I've not started using interactive mode yet. The first time I tried it, rasa had to retrain and I got pissed off !
Takes about 2 hours for me to complete training