OpenHermes-2.5-Mistral-7B-Sentence-Segmentation

See files for original notebook used for finetuning

Model description

OpenHermes-2.5-Mistral-7B-Sentence-Segmentation is a DPO finetuned OpenHermes model for sentence segmentation capability.

This model is based on teknium/OpenHermes-2.5-Mistral-7B, a state-of-the-art chat-aligned 7B model.

Example Outputs

The model has been finetuned with (ChatML)[https://github.com/MicrosoftDocs/azure-docs/blob/main/articles/ai-services/openai/includes/chat-markup-language.md#messages] template:

<|im_start|>system
Segment:<|im_end|>
<|im_start|>user
```yaml
"input":
  "sentence":
    "segment":
    - "word": "Shere,"
    - "word": "in"
    - "word": "your"
    - "word": "report"
    - "word": "on"
    - "word": "female"
    - "word": "sexuality,"
    - "word": "men"
    - "word": "were"
    - "word": "staggered"
    - "word": "to"
    - "word": "learn"
    - "word": "that"
    - "word": "clitoral"
    - "word": "stimulation"
    - "word": "was"
    - "word": "much"
    - "word": "more"
    - "word": "important"
    - "word": "than"
    - "word": "penetration."
```<|im_end|>
<|im_start|>assistant

with output in the format of

```yaml
"output":
  "sentence":
    "segment":
    - "word": "Shere,"
    - "word": "in"
    - "word": "your"
    - "word": "report"
    - "word": "on"
    - "word": "female"
    - "word": "sexuality,"
    "segment":
    - "word": "men"
    - "word": "were"
    - "word": "staggered"
    - "word": "to"
    - "word": "learn"
    - "word": "that"
    "segment":
    - "word": "clitoral"
    - "word": "stimulation"
    - "word": "was"
    - "word": "much"
    - "word": "more"
    - "word": "important"
    - "word": "than"
    - "word": "penetration."

## Misc

This mistral model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.

[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)