Update README.md

README.md

@@ -1,17 +1,29 @@
 ---
 library_name: transformers
-model_name: Llama-3.1-Argunaut-1-8B-SPIN-dev4
+model_name: Llama-3.1-Argunaut-1-8B-SPIN
+pipeline_tag: text-generation
+base_model: DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT
+datasets:
+- DebateLabKIT/argdown_line-by-line
+- DebateLabKIT/argument_mapping_dpo_pairs
+- allenai/llama-3.1-tulu-3-70b-preference-mixture
 tags:
+- logic
+- argumentation
+- critical-thinking
+- argument-mapping
 - generated_from_trainer
 - trl
 - dpo
-licence: license
+- spin
+licence: llama3.1
 ---
 
-# Model Card for Llama-3.1-Argunaut-1-8B-SPIN-dev4
+# Model Card for Llama-3.1-Argunaut-1-8B-SPIN
 
-This model is a fine-tuned version of [None](https://huggingface.co/None).
-It has been trained using [TRL](https://github.com/huggingface/trl).
+
+This model is a fine-tuned version of [DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT](https://huggingface.co/DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT).
+It has been trained using [TRL](https://github.com/huggingface/trl) and [vLLM](https://docs.vllm.ai/). Checkpoints are tagged.
 
 ## Quick start
 
@@ -19,17 +31,18 @@ It has been trained using [TRL](https://github.com/huggingface/trl).
 from transformers import pipeline
 
 question = "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?"
-generator = pipeline("text-generation", model="DebateLabKIT/Llama-3.1-Argunaut-1-8B-SPIN-dev4", device="cuda")
+generator = pipeline("text-generation", model="DebateLabKIT/Llama-3.1-Argunaut-1-8B-SPIN", device="cuda")
 output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
 print(output["generated_text"])
 ```
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/ggbetz/argunauts-training/runs/d55470xf) 
+<!--[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/ggbetz/argunauts-training/runs/s89n820x)-->
 
+This model was trained with Self-Play Fine-Tuning (SPIN), a method introduced in [Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models](https://huggingface.co/papers/2401.01335).
 
-This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
+More details about the training procedure will be released in a blog post!
 
 ### Framework versions
 
@@ -41,16 +54,17 @@ This model was trained with DPO, a method introduced in [Direct Preference Optim
 
 ## Citations
 
-Cite DPO as:
+Cite SPIN as:
 
 ```bibtex
-@inproceedings{rafailov2023direct,
-    title        = {{Direct Preference Optimization: Your Language Model is Secretly a Reward Model}},
-    author       = {Rafael Rafailov and Archit Sharma and Eric Mitchell and Christopher D. Manning and Stefano Ermon and Chelsea Finn},
-    year         = 2023,
-    booktitle    = {Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023},
-    url          = {http://papers.nips.cc/paper_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html},
-    editor       = {Alice Oh and Tristan Naumann and Amir Globerson and Kate Saenko and Moritz Hardt and Sergey Levine},
+@misc{chen2024selfplayfinetuningconvertsweak,
+      title={Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models}, 
+      author={Zixiang Chen and Yihe Deng and Huizhuo Yuan and Kaixuan Ji and Quanquan Gu},
+      year={2024},
+      eprint={2401.01335},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2401.01335}, 
 }
 ```