Update README.md

README.md

@@ -10,52 +10,126 @@ model-index:
   results: []
 ---
 
-<!-- This model card has been generated automatically according to the information the Trainer had access to. You
-should probably proofread and complete it, then remove this comment. -->
+# gemma-2-9b-it-SimPO Model Card
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](None)
-# gemma2-9b-simpo-beta-10-ratio-0.4-lr-8e-7
+SimPO (Simple Preference Optimization) is an offline preference optimization algorithm designed to enhance the training of large language models (LLMs) with preference optimization datasets. SimPO aligns the reward function with the generation likelihood, eliminating the need for a reference model and incorporating a target reward margin to boost performance. Please refer to our [preprint](https://arxiv.org/pdf/2405.14734) and [github repo](https://github.com/princeton-nlp/SimPO) for more details.
 
-This model is a fine-tuned version of [/scratch/gpfs/DANQIC/ym0081/hf_cache/gemma-2-9b-it](https://huggingface.co//scratch/gpfs/DANQIC/ym0081/hf_cache/gemma-2-9b-it) on the /scratch/gpfs/DANQIC/ym0081/hf_cache/gemma2-ultrafeedback-armorm/dataset_dict/ dataset.
 
-## Model description
+## Model Details
 
-More information needed
+### Model Description
 
-## Intended uses & limitations
+We fine-tuned [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on [princeton-nlp/gemma2-ultrafeedback-armorm](https://huggingface.co/datasets/princeton-nlp/gemma2-ultrafeedback-armorm) with the SimPO objective.
 
-More information needed
+- **Developed by:** Yu Meng, Mengzhou Xia, Danqi Chen
+- **Model type:** Causal Language Model
+- **License:** gemma
+- **Finetuned from model:** [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it)
 
-## Training and evaluation data
+### Model Sources
 
-More information needed
+<!-- Provide the basic links for the model. -->
 
-## Training procedure
+- **Repository:** https://github.com/princeton-nlp/SimPO
+- **Paper:** https://arxiv.org/pdf/2405.14734
+- **Demo:** Soon to be alive
 
-### Training hyperparameters
 
-The following hyperparameters were used during training:
-- learning_rate: 8e-07
-- train_batch_size: 2
-- eval_batch_size: 4
-- seed: 42
-- distributed_type: multi-GPU
-- num_devices: 8
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 128
-- total_eval_batch_size: 32
-- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
-- lr_scheduler_type: cosine
-- lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 1
+## How to Get Started with the Model
+```
+import torch
+from transformers import pipeline
 
-### Training results
+model_id = "princeton-nlp/gemma-2-9b-it-SimPO"
 
+generator = pipeline(
+    "text-generation",
+    model=model_id,
+    model_kwargs={"torch_dtype": torch.bfloat16},
+    device="cuda",
+)
+outputs = generator([{"role": "user", "content": "What's the difference between llamas and alpacas?"}], do_sample=False, max_new_tokens=200)
+print(outputs[0]['generated_text'])
+```
 
+## Training Details
 
-### Framework versions
+### Training Data
 
-- Transformers 4.42.4
-- Pytorch 2.2.2+cu121
-- Datasets 2.18.0
-- Tokenizers 0.19.1
+We use [princeton-nlp/gemma2-ultrafeedback-armorm](https://huggingface.co/datasets/princeton-nlp/gemma2-ultrafeedback-armorm) as the preference optimization dataset.
+
+#### Training Hyperparameters
+
+[TO BE FILLED LATER]
+
+#### Speeds, Sizes, Times
+
+Fine-tuning the [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on [princeton-nlp/gemma2-ultrafeedback-armorm](https://huggingface.co/datasets/princeton-nlp/gemma2-ultrafeedback-armorm) takes around 100 mins to finish on 8xH100 GPUs.
+
+## Evaluation Results
+
+
+|               models                    | AE2 LC | AE2 WR | AE2 Length |  AH  | AH Length |  GSM | GSM Length | MMLU | MMLU Length |
+|-----------------------------------|:------:|:------:|:----------:|:----:|:---------:|:----:|:----------:|:----:|:-----------:|
+|        [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it)       |  51.1  |  38.1  |    1571    | 40.8 |    545    | 87.4 |     395    | 72.7 |     515     |
+|  [princeton-nlp/gemma-2-9b-it-DPO](https://huggingface.co/princeton-nlp/gemma-2-9b-it-DPO)  |  67.8  |  65.4  |    2016    | 58.9 |    717    | 88.5 |     392    | 72.2 |     624     |
+| [princeton-nlp/gemma-2-9b-it-SimPO](https://huggingface.co/princeton-nlp/gemma-2-9b-it-SimPO) |  72.4  |  65.9  |    1833    | 59.1 |    693    | 88.0 |     341    | 72.2 |     441     |
+
+
+## Technical Specifications
+
+### Model Architecture and Objective
+
+The model architecture is based on [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it). We use the SimPO training objective proposed in our [preprint](https://arxiv.org/pdf/2405.14734).
+
+#### Hardware
+
+We used 8xH100 GPUs for model training.
+
+#### Software
+
+Training was done using the [alignment-handbook](https://github.com/huggingface/alignment-handbook) library.
+
+## Citation
+
+gemma model:
+```
+@article{gemma_2024,
+    title={Gemma},
+    url={https://www.kaggle.com/m/3301},
+    DOI={10.34740/KAGGLE/M/3301},
+    publisher={Kaggle},
+    author={Gemma Team},
+    year={2024}
+}
+```
+
+SimPO paper:
+```
+@article{meng2024simpo,
+  title={{SimPO}: Simple preference optimization with a reference-free reward},
+  author={Meng, Yu and Xia, Mengzhou and Chen, Danqi},
+  journal={arXiv preprint arXiv:2405.14734},
+  year={2024}
+}
+```
+
+UltraFeedback paper:
+```
+@article{cui2023ultrafeedback,
+  title={{UltraFeedback}: Boosting language models with high-quality feedback},
+  author={Cui, Ganqu and Yuan, Lifan and Ding, Ning and Yao, Guanming and Zhu, Wei and Ni, Yuan and Xie, Guotong and Liu, Zhiyuan and Sun, Maosong},
+  journal={arXiv preprint arXiv:2310.01377},
+  year={2023}
+}
+```
+
+ArmoRM paper:
+```
+@article{wang2024interpretable,
+  title={Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts},
+  author={Wang, Haoxiang and Xiong, Wei and Xie, Tengyang and Zhao, Han and Zhang, Tong},
+  journal={arXiv preprint arXiv:2406.12845},
+  year={2024}
+}
+```