---
language: 
  - en
tags:
  - NLP
license: mit
datasets:
  - TristanBehrens/bach_garland_2024-100K
base_model: None
---

# bach_garland_transformer - An xLSTM Model

![Trained with Helibrunna](banner.jpg)

Trained with [Helibrunna](https://github.com/AI-Guru/helibrunna) by [Dr. Tristan Behrens](https://de.linkedin.com/in/dr-tristan-behrens-734967a2).

## Configuration

```
training:
  model_name: bach_garland_transformer
  batch_size: 40
  lr: 0.001
  lr_warmup_steps: 1000
  lr_decay_until_steps: 10000
  lr_decay_factor: 0.001
  weight_decay: 0.1
  amp_precision: bfloat16
  weight_precision: float32
  enable_mixed_precision: true
  num_epochs: 8
  output_dir: output/bach_garland_transformer
  save_every_step: 500
  log_every_step: 10
  wandb_project: bach_garland
  torch_compile: false
model:
  type: transformer
  dim: 64
  n_layers: 4
  n_heads: 4
  fc_scale: 2
  context_length: 2048
  vocab_size: 178
dataset:
  hugging_face_id: TristanBehrens/bach_garland_2024-100K
tokenizer:
  type: whitespace
  fill_token: '[EOS]'

```