---
base_model: unsloth/mistral-7b-instruct-v0.3-bnb-4bit
library_name: peft
license: gpl-3.0
datasets:
- ArsParadox/Viel_Dataset_Lite
language:
- en
pipeline_tag: text-generation
tags:
- rp
---

# Model Card for Model ID

This is a Version 3 of Viel Model

It's an experiment of creating an AI with built-in personality. 

An upgrade from Viel v2, this one actually uses proper dataset formatting

Also uses Mistral instead of Llama 3 because we want personality, not IQ


![image/png](https://cdn-uploads.huggingface.co/production/uploads/667d8f82eda09e9a8b9994aa/z-199xmSUBUPbLxm0GGHl.png)

OH MY GOD IT WORKS!!!


ALSO: Quants Here: https://huggingface.co/mradermacher/Viel-Mistral-v3-GGUF


## Model Details

### Character Detail

Viel, an industrial grade robot repurposed as shitty assistant

### Model Description

- **Developed by:** Ars Paradox
- **Funded by [optional]:** Google Colab
- **Model type:** Mistral 7B Instruct
- **License:** GPL-3
- **Finetuned from model [optional]:** Mistral 7B Instruct

## Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

Use the chat_ml format to run the model. 

No need to add any additional instruction. Just start talking. You'll see how it works.

### Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Viel is inaccurate not that smart and she knows it


## Training Details

### Training Data

ArsParadox/Viel_Dataset_Lite

### Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Unsloth...

Uhhh...

```
from trl import SFTTrainer
from transformers import TrainingArguments
from unsloth import is_bfloat16_supported

trainer = SFTTrainer(
    model = model,
    tokenizer = tokenizer,
    train_dataset = dataset,
    dataset_text_field = "text",
    max_seq_length = max_seq_length,
    dataset_num_proc = 2,
    packing = False, # Can make training 5x faster for short sequences.
    args = TrainingArguments(
        per_device_train_batch_size = 2,
        gradient_accumulation_steps = 4,
        warmup_steps = 5,
        # num_train_epochs=5,
        max_steps = 180,
        learning_rate = 2e-4,
        fp16 = not is_bfloat16_supported(),
        bf16 = is_bfloat16_supported(),
        logging_steps = 1,
        optim = "adamw_8bit",
        weight_decay = 0.01,
        lr_scheduler_type = "linear",
        seed = 3407,
        output_dir = "outputs",
        report_to = "none", # Use this for WandB etc
    ),
)
```
Does that answer your question?


## Model Card Contact

Discord @pandu.paradox for further query.

Feel free to test it, see how it works~

Hopefully this time, the personality gets embedded better than the last 3 model.

I will work on another character next.