|
--- |
|
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`](https://huggingface.co/julep-ai/samantha-33b/blob/main/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) |
|
``` |
