Quantization made by Richard Erkhov.

[Github](https://github.com/RichardErkhov)

[Discord](https://discord.gg/pvy7H8DZMG)

[Request more models](https://github.com/RichardErkhov/quant_request)


AID-Neo-125M - bnb 8bits
- Model creator: https://huggingface.co/Crataco/
- Original model: https://huggingface.co/Crataco/AID-Neo-125M/




Original model description:
---
language: en
license: mit
pipeline_tag: text-generation
---

# **UPDATE (2023-09-23):**
This model is obsolete. Thanks to quantization you can run AI Dungeon 2 Classic (a 1.5B model) under equivalent hardware. [See here](https://huggingface.co/Crataco/ggml-ai-dungeon-2-classic).
***
# AID-Neo-125M

## Model description
This model was inspired by -- and finetuned on the same dataset of -- [KoboldAI's GPT-Neo-125M-AID (Mia) model](https://huggingface.co/KoboldAI/GPT-Neo-125M-AID): the AI Dungeon dataset (`text_adventures.txt`). This was to fix a possible oversight in the original model, which was trained with [an unfortunate bug](https://github.com/EricFillion/happy-transformer/issues/283). You could technically consider it a "retraining" of the same model using different software.

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0