---
library_name: transformers
license: apache-2.0
language:
- de
datasets:
- devngho/culturax-mini-nonshuffled
- maxidl/FineNews-unfiltered
- djstrong/oscar-small
- LemiSt/gutenberg_de
- almanach/HALvest
- wikimedia/wikipedia
- D4ve-R/terra-xplain-cc-de
base_model:
- HuggingFaceTB/SmolLM-135M
pipeline_tag: text-generation
---

# Model Card for SmolLM-135M-de

A german version of [HuggingFaceTB/SmolLM-135M](https://huggingface.co/HuggingFaceTB/SmolLM-135M/blob/main/README.md), trained to speak German by applying CPT for about 6 billion tokens.


If you are looking for a chat model, try [this](https://huggingface.co/LemiSt/SmolLM-135M-instruct-de-merged) fine tune or the [corresponding adapter model](https://huggingface.co/LemiSt/SmolLM-135M-instruct-de).

## Model Details

### Model Description

The base model is [HuggingFaceTB/SmolLM-135M](https://huggingface.co/HuggingFaceTB/SmolLM-135M/blob/main/README.md), which I further trained on about 6 billion German-language tokens.

- **Model type:** Large Language Model (Llama architecture)
- **Language(s) (NLP):** German
- **License:** Apache 2.0
- **Finetuned from model:** [HuggingFaceTB/SmolLM-135M](https://huggingface.co/HuggingFaceTB/SmolLM-135M/blob/main/README.md)

## Uses

I mainly made this as a small experimentation model to quickly benchmark datasets etc. - since the model is so small, I am unsure about its usefulness for any real-world scenarios.

This is a base model without any chat fine tuning etc. and thus should not be used as-is. It outputs mostly correct German, which is what I tried to achieve.

If you are looking for a chat model, try [this](https://huggingface.co/LemiSt/SmolLM-135M-instruct-de) adapter. 

## Bias, Risks, and Limitations

This is a very small model and will output blatantly wrong information. I have not done any further filtering on the source datasets, so it is possible that the model will generate lewd or otherwise inappropriate content. Use with care.

I would **strongly** recommend against using this model in a production setting, at least without further fine tuning and preference optimization.

## How to Get Started with the Model

Use the code below to get started with the model.

```python
# adapted from the original SmolLM repo
# pip install transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "LemiSt/SmolLM-135M-de"
device = "cuda" # for GPU usage or "cpu" for CPU usage
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
# for multiple GPUs install accelerate and do `model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map="auto")`
model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device)
inputs = tokenizer.encode("Rezept für einen leckeren veganen Schokokuchen:\n", return_tensors="pt").to(device)
outputs = model.generate(inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0]))
```

## Training Details

### Training Data

  - [devngho/culturax-mini-nonshuffled](https://huggingface.co/datasets/devngho/culturax-mini-nonshuffled)
  - [maxidl/FineNews-unfiltered](https://huggingface.co/datasets/maxidl/FineNews-unfiltered) CC-NEWS-2024-05 config, de split
  - [djstrong/oscar-small](https://huggingface.co/datasets/djstrong/oscar-small) unshuffled_deduplicated_de config
  - [LemiSt/gutenberg_de](https://huggingface.co/datasets/LemiSt/gutenberg_de)
  - [almanach/HALvest](https://huggingface.co/datasets/almanach/HALvest) de config
  - [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) 20231101.de config
  - [D4ve-R/terra-xplain-cc-de](https://huggingface.co/datasets/D4ve-R/terra-xplain-cc-de)

### Training Procedure

This was trained with axolotl, using full fine tuning (no LoRA etc). I used a sequence length of 2048 with an effective batch size of 512, learning rate of 0.003 with the adamw_bnb_8bit optimizer and a cosine scheduler.
Due to an error I made in calculating the token count, I accidentally trained for nearly 2 epochs, with the learning rate not reaching its proper minimum.