You can deploy this model and start earning money today!

RichardErkhov/
UCLA-AGI_-_Gemma-2-9B-It-SPPO-Iter3-gguf

Name: UCLA-AGI_-_Gemma-2-9B-It-SPPO-Iter3-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Gemma-2-9B-It-SPPO-Iter3 - GGUF

Model creator: https://huggingface.co/UCLA-AGI/
Original model: https://huggingface.co/UCLA-AGI/Gemma-2-9B-It-SPPO-Iter3/

Name	Quant method	Size
Gemma-2-9B-It-SPPO-Iter3.Q2_K.gguf	Q2_K	3.54GB
Gemma-2-9B-It-SPPO-Iter3.IQ3_XS.gguf	IQ3_XS	3.86GB
Gemma-2-9B-It-SPPO-Iter3.IQ3_S.gguf	IQ3_S	4.04GB
Gemma-2-9B-It-SPPO-Iter3.Q3_K_S.gguf	Q3_K_S	4.04GB
Gemma-2-9B-It-SPPO-Iter3.IQ3_M.gguf	IQ3_M	4.19GB
Gemma-2-9B-It-SPPO-Iter3.Q3_K.gguf	Q3_K	4.43GB
Gemma-2-9B-It-SPPO-Iter3.Q3_K_M.gguf	Q3_K_M	4.43GB
Gemma-2-9B-It-SPPO-Iter3.Q3_K_L.gguf	Q3_K_L	4.78GB
Gemma-2-9B-It-SPPO-Iter3.IQ4_XS.gguf	IQ4_XS	4.86GB
Gemma-2-9B-It-SPPO-Iter3.Q4_0.gguf	Q4_0	5.07GB
Gemma-2-9B-It-SPPO-Iter3.IQ4_NL.gguf	IQ4_NL	5.1GB
Gemma-2-9B-It-SPPO-Iter3.Q4_K_S.gguf	Q4_K_S	5.1GB
Gemma-2-9B-It-SPPO-Iter3.Q4_K.gguf	Q4_K	5.37GB
Gemma-2-9B-It-SPPO-Iter3.Q4_K_M.gguf	Q4_K_M	5.37GB
Gemma-2-9B-It-SPPO-Iter3.Q4_1.gguf	Q4_1	5.55GB
Gemma-2-9B-It-SPPO-Iter3.Q5_0.gguf	Q5_0	6.04GB
Gemma-2-9B-It-SPPO-Iter3.Q5_K_S.gguf	Q5_K_S	6.04GB
Gemma-2-9B-It-SPPO-Iter3.Q5_K.gguf	Q5_K	6.19GB
Gemma-2-9B-It-SPPO-Iter3.Q5_K_M.gguf	Q5_K_M	6.19GB
Gemma-2-9B-It-SPPO-Iter3.Q5_1.gguf	Q5_1	6.52GB
Gemma-2-9B-It-SPPO-Iter3.Q6_K.gguf	Q6_K	7.07GB
Gemma-2-9B-It-SPPO-Iter3.Q8_0.gguf	Q8_0	9.15GB

Original model description:

license: gemma datasets:

openbmb/UltraFeedback language:
en pipeline_tag: text-generation

Self-Play Preference Optimization for Language Model Alignment (https://arxiv.org/abs/2405.00675)

Gemma-2-9B-It-SPPO-Iter3

This model was developed using Self-Play Preference Optimization at iteration 3, based on the google/gemma-2-9b-it architecture as starting point. We utilized the prompt sets from the openbmb/UltraFeedback dataset, splited to 3 parts for 3 iterations by snorkelai/Snorkel-Mistral-PairRM-DPO-Dataset. All responses used are synthetic.

Terms of Use: Terms

Links to Other Models

Model Description

Model type: A 8B parameter GPT-like model fine-tuned on synthetic datasets.
Language(s) (NLP): Primarily English
License: Apache-2.0
Finetuned from model: google/gemma-2-9b-it

AlpacaEval Leaderboard Evaluation Results

Model	LC. Win Rate	Win Rate	Avg. Length
Gemma-2-9B-SPPO Iter1	48.70	40.76	1669
Gemma-2-9B-SPPO Iter2	50.93	44.64	1759
Gemma-2-9B-SPPO Iter3	53.27	47.74	1803

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-07
eta: 1000
per_device_train_batch_size: 8
gradient_accumulation_steps: 1
seed: 42
distributed_type: deepspeed_zero3
num_devices: 8
optimizer: RMSProp
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_train_epochs: 1.0

Citation

@misc{wu2024self,
      title={Self-Play Preference Optimization for Language Model Alignment}, 
      author={Wu, Yue and Sun, Zhiqing and Yuan, Huizhuo and Ji, Kaixuan and Yang, Yiming and Gu, Quanquan},
      year={2024},
      eprint={2405.00675},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

RichardErkhov/UCLA-AGI_-_Gemma-2-9B-It-SPPO-Iter3-gguf