IKEA-7b-UC-0

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0241
• Rewards/chosen: -1.3913
• Rewards/rejected: -26.2733
• Rewards/accuracies: 0.9940
• Rewards/margins: 24.8821
• Logps/rejected: -348.4335
• Logps/chosen: -162.1273
• Logits/rejected: -2.3713
• Logits/chosen: -2.5987

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-07
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 16
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.0268	0.72	500	0.0314	-0.9700	-9.7922	0.9960	8.8222	-183.6220	-157.9143	-2.2332	-2.4086
0.0022	1.44	1000	0.0229	-1.6108	-24.5526	0.9960	22.9418	-331.2263	-164.3232	-2.4361	-2.6330

Framework versions

• Transformers 4.36.1
• Pytorch 2.0.1+cu117
• Datasets 2.16.1
• Tokenizers 0.15.0