metadata

license: apache-2.0
library_name: peft
tags:
  - trl
  - dpo
  - generated_from_trainer
base_model: mistralai/Mistral-7B-v0.1
model-index:
  - name: zephyr-7b
    results: []

zephyr-7b

This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.5769
Rewards/chosen: -0.6646
Rewards/rejected: -1.1353
Rewards/accuracies: 0.3711
Rewards/margins: 0.4707
Logps/rejected: -190.7267
Logps/chosen: -130.3719
Logits/rejected: 1.8500
Logits/chosen: 1.7576
Use Label: 6517.1875
Pred Label: 782.8125

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-06
train_batch_size: 4
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 4
total_train_batch_size: 128
total_eval_batch_size: 64
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

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen	Use Label	Pred Label
0.6531	0.21	100	0.6528	-0.1643	-0.2945	0.3633	0.1303	-106.6470	-80.3385	-1.7198	-1.7354	1725.3125	6.6875
0.6041	0.42	200	0.5936	-0.7144	-1.1047	0.3516	0.3903	-187.6596	-135.3474	0.9784	0.8864	3420.5938	167.4062
0.5763	0.63	300	0.5773	-0.7930	-1.2317	0.3516	0.4387	-200.3615	-143.2137	1.7526	1.6599	4991.2812	452.7188
0.5836	0.84	400	0.5769	-0.6646	-1.1353	0.3711	0.4707	-190.7267	-130.3719	1.8500	1.7576	6517.1875	782.8125

Framework versions

PEFT 0.7.1
Transformers 4.38.2
Pytorch 2.1.1+cu121
Datasets 2.14.6
Tokenizers 0.15.2