samantha-1-alpha / README.md
diwank's picture
Upload tokenizer
7ea6747 verified
metadata
language:
  - en
license: mit
library_name: transformers
datasets:
  - julep-ai/samantha_finetune_dataset_03
pipeline_tag: text-generation

Samantha

Technical notes

This model is trained on a specialized dataset and uses special sentinel tokens to demarcate conversations.

Usage

For usage, you can refer to the chat.py file in this repo for an example.

Concepts

  • Each conversation consists of n "sections"
  • Each section can be one of:
    • me: The model
    • person: The speaker
    • situation: relevant background information to set the context of the conversation
    • thought: Thoughts generated by the model for parsing intermediate steps etc
    • information: External information added into the context by the system running the model
  • The model and speaker sections can optionally include a name like me (Samantha) or person (Dmitry)

Sentinel Tokens

  • <|im_start|> token marks the start of a "section"
  • <|im_end|> token marks the end of a "section".

Example

<|im_start|>situation
I am talking to Diwank. I want to ask him about his food preferences.<|im_end|>
<|im_start|>person (Diwank)
Hey Samantha! What do you want to talk about?<|im_end|>
<|im_start|>me (Samantha)