---
license: llama3.3
datasets:
- DebateLabKIT/deepa2-conversations
- DebateLabKIT/deep-argmap-conversations
- allenai/tulu-3-sft-mixture
base_model:
- meta-llama/Llama-3.3-70B-Instruct
pipeline_tag: text-generation
library_name: transformers
tags:
- logic
- argumentation
- critical-thinking
- argument-mapping
- trl
- sft
---

# Model Card for Llama-3.3-Argunaut-1-70B-SFT

This model is a fine-tuned version of [meta-llama/Llama-3.3-70B-Instruct](https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct).
It has been trained using [TRL](https://github.com/huggingface/trl).

## Quick start

```python
from transformers import pipeline

question = "Are you familiar with Argdown syntax? What's its purpose?"
generator = pipeline("text-generation", model="DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```


## SFT dataset mixture

|Dataset|Weight (examples)|Weight (tokens)|
|:------|:----:|:----:|
|DebateLabKIT/deepa2-conversations|25%|49%|
|DebateLabKIT/deep-argmap-conversations|25%|18%|
|allenai/tulu-3-sft-mixture|50%|33%|


## Training procedure

Trained with SFT on **1M examples** and for 1 epoch with 

* context length 8196
* packing (trl implementation)
* *spectrum* (top 30 percent)

```yaml
# Training parameters
num_train_epochs: 1
per_device_train_batch_size: 2
gradient_accumulation_steps: 8
gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
learning_rate: 2.0e-6  # following _Tülu 3_ recipe
lr_scheduler_type: cosine
warmup_ratio: 0.1
```

Hardware: 4 x H100 GPUs.

_This work was performed on the HoreKa supercomputer funded by the
Ministry of Science, Research and the Arts Baden-Württemberg and by
the Federal Ministry of Education and Research._

### Framework versions

- TRL: 0.12.1
- Transformers: 4.46.3
- Pytorch: 2.4.1
- Datasets: 3.1.0
- Tokenizers: 0.20.3

## Credits 

This work wouldn't be possible without all the **great contributions from the open LLM community**. Thank you! Special kudos go to 

- @philschmid for his latest [fine-tuning boilerplate](https://www.philschmid.de/fine-tune-llms-in-2025)
- @lvwerra, @lewtun et al for building and maintaining [trl](https://github.com/huggingface/trl)
- @cognitivecomputations for sharing [spectrum](https://github.com/cognitivecomputations/spectrum/tree/main)
- @allenai for the [Tülu recipe and artifacts](https://huggingface.co/collections/allenai/tulu-3-datasets-673b8df14442393f7213f372)