README.md

---
license: apache-2.0
---

# LimaRP-Mistral-7B-v0.1 (Alpaca, 8-bit LoRA adapter)

This is a version of LimaRP for [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) with
about 1800 training samples _up to_ 4k tokens length. A 2-pass training procedure has been employed. The first pass includes
finetuning on about 6800 stories within 4k tokens length and the second pass is LimaRP with changes introducing more effective
control on response length.

**Due to software limitations, finetuning didn't take advantage yet of the Sliding Window Attention (SWA) which would have allowed
to use longer conversations in the training data. Thus, this version of LimaRP should be considered an _initial finetuning attempt_ and
will be updated in the future.**

For more details about LimaRP, see the model page for the [previously released v2 version for Llama-2](https://huggingface.co/lemonilia/limarp-llama2-v2).
Most details written there apply for this version as well. Generally speaking, LimaRP is a longform-oriented, novel-style
roleplaying chat model intended to replicate the experience of 1-on-1 roleplay on Internet forums. Short-form,
IRC/Discord-style RP (aka "Markdown format") is not supported yet. The model does not include instruction tuning,
only manually picked and slightly edited RP conversations with persona and scenario data.

## Prompt format
Same as before. It uses the [extended Alpaca format](https://github.com/tatsu-lab/stanford_alpaca),
with `### Input:` immediately preceding user inputs and `### Response:` immediately preceding
model outputs. While Alpaca wasn't originally intended for multi-turn responses, in practice this
is not a problem; the format follows a pattern already used by other models.

```
### Instruction:
Character's Persona: {bot character description}

User's Persona: {user character description}

Scenario: {what happens in the story}

Play the role of Character. You must engage in a roleplaying chat with User below this line. Do not write dialogues and narration for User.

### Input:
User: {utterance}

### Response:
Character: {utterance}

### Input
User: {utterance}

### Response:
Character: {utterance}

(etc.)
```

You should:
- Replace all text in curly braces (curly braces included) with your own text.
- Replace `User` and `Character` with appropriate names.


### Message length control
Inspired by the previously named "Roleplay" preset in SillyTavern, with this
version of LimaRP it is possible to append a length modifier to the response instruction
sequence, like this:

```
### Input
User: {utterance}

### Response: (length = medium)
Character: {utterance}
```

This has an immediately noticeable effect on bot responses. The available lengths are:
`tiny`, `short`, `medium`, `long`, `huge`, `humongous`, `extreme`, `unlimited`. **The 
recommended starting length is `medium`**. Keep in mind that the AI may ramble
or impersonate the user with very long messages.

The length control effect is reproducible, but the messages will not necessarily follow
lengths very precisely, rather follow certain ranges on average, as seen in this table
with data from tests made with one reply at the beginning of the conversation:

![lengths](https://files.catbox.moe/dy39bt.png)

Response length control appears to work well also deep into the conversation.

## Suggested settings
You can follow these instruction format settings in SillyTavern. Replace `tiny` with
your desired response length:

![settings](https://files.catbox.moe/6lcz0u.png)

## Text generation settings
Extensive testing with Mistral has not been performed yet, but suggested starting text
generation settings may be:

- TFS = 0.90~0.95
- Temperature = 0.70~0.85
- Repetition penalty = 1.08~1.10
- top-k = 0 (disabled)
- top-p = 1 (disabled)

## Training procedure
[Axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) was used for training
on a 4x NVidia A40 GPU cluster.

The A40 GPU cluster has been graciously provided by [Arc Compute](https://www.arccompute.io/).

The model has been trained as an 8-bit LoRA adapter, and
it's so large because a LoRA rank of 256 was also used. The reasoning was that this
might have helped the model internalize any newly acquired information, making the
training process closer to a full finetune. It's suggested to merge the adapter to
the base Mistral-7B-v0.1 model.

### Training hyperparameters
- learning_rate: 0.0001
- lr_scheduler_type: cosine
- num_epochs: 2 (1 for the first pass)
- sequence_len: 4096
- lora_r: 256
- lora_alpha: 16
- lora_dropout: 0.05
- lora_target_linear: True
- bf16: True
- fp16: false
- tf32: True
- load_in_8bit: True
- adapter: lora
- micro_batch_size: 2
- gradient_accumulation_steps: 1
- warmup_steps: 40
- optimizer: adamw_torch

For the second pass, the `lora_model_dir` option was used to continue finetuning on the LoRA
adapter obtained from the first pass.

Using 2 GPUs, the effective global batch size would have been 8.