README.md

---
license: apache-2.0
tags:
- audio
- speech
- language-model
- auristream
library_name: transformers
---

# AuriStream100M_80Pred_BigAudioDataset_500k

**AuriStream** is a speech language model by **Greta Tuckute** and **Klemen Kotar**.

This model predicts cochlear tokens from a tokenizer such as [WavCochCausalV8192](https://huggingface.co/TuKoResearch/WavCochCausalV8192).



## Model Details

| Parameter | Value |
|-----------|-------|
| Parameters | ~0.59B |
| Layers | 12 |
| Hidden Size | 768 |
| Attention Heads | 12 |
| Vocab Size | 8192 |
| Prediction Steps | 80 |

## Usage

```python
from transformers import AutoModel, AutoConfig

# Load with trust_remote_code for custom model
model = AutoModel.from_pretrained(
    "TuKoResearch/AuriStream100M_80Pred_BigAudioDataset_500k",
    trust_remote_code=True,
)

# Or load config first
config = AutoConfig.from_pretrained("TuKoResearch/AuriStream100M_80Pred_BigAudioDataset_500k", trust_remote_code=True)
```

## Base Model Code

This checkpoint uses shared model code from [TuKoResearch/AuriStream-base](https://huggingface.co/TuKoResearch/AuriStream-base).

## Tokenizer

This model uses cochlear tokens from [WavCochCausalV8192](https://huggingface.co/TuKoResearch/WavCochCausalV8192).