metadata
datasets:
- Kamisori-daijin/email-datasets-20k
language:
- en
pipeline_tag: text-generation
tags:
- small
- cpu
- email
- mail
- e-mail
- tiny
- nanogpt
- gpt
- open
- opensource
Welcome to eMail-LM
This is a 44.64M parameter GPT model which generates E-Mails.
How to use
To use this model, download model.pt and use.py from this repo and run use.py.
Sample
Input: Write an email to my boss about a server outage.
Output:
SUBJECT: Regarding the Server Outage - Seriously?
BODY:
[Name],
Let's be blunt. The server went down last night, and now a bad timing – on a Friday night. Frankly, it was unacceptable. I'm not going to sugarcoat it.
I'm not interested in excuses about how this happened, and I'm fixing it. I expect a full report by [Time] today.
[Your Name]
Training code
The full training code can be found as train.ipynb in this repo. Have fun :D
Limitations
As this is a very small model and it wasn't trained for a really long time, it still generates garbage - but always emails. It's definetely more for research purposes.