moos124/sdft-model

1---2base_model: Qwen/Qwen2.5-1.5B-Instruct3library_name: transformers4model_name: sdft-model5tags:6- generated_from_trainer7- trl8- sdft9- hf_jobs10licence: license11---12 13# Model Card for sdft-model14 15This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct).16It has been trained using [TRL](https://github.com/huggingface/trl).17 18## Quick start19 20```python21from transformers import pipeline22 23question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"24generator = pipeline("text-generation", model="moos124/sdft-model", device="cuda")25output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]26print(output["generated_text"])27```28 29## Training procedure30 31 32 33 34 35This model was trained with SDFT, a method introduced in [Self-Training with On-Policy Self-Distillation for Language Model Alignment](https://huggingface.co/papers/2601.19897).36 37### Framework versions38 39- TRL: 1.3.040- Transformers: 5.7.041- Pytorch: 2.11.042- Datasets: 4.8.543- Tokenizers: 0.22.244 45## Citations46 47Cite SDFT as:48 49```bibtex50@article{hubotter2026selftraining,51    title        = {{Self-Training with On-Policy Self-Distillation for Language Model Alignment}},52    author       = {Jonas H\"ubotter and Frederike L\"ubeck and Lejs Behric and Anton Baumann and Marco Bagatella and Daniel Marta and Ido Hakimi and Idan Shenfeld and Thomas Kleine Buening and Carlos Guestrin and Andreas Krause},53    year         = 2026,54    eprint       = {arXiv:2601.19897}55}56```57 58Cite TRL as:59    60```bibtex61@software{vonwerra2020trl,62  title   = {{TRL: Transformers Reinforcement Learning}},63  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},64  license = {Apache-2.0},65  url     = {https://github.com/huggingface/trl},66  year    = {2020}67}68```