zephyr0-7b-ultra-p-0.05

This model is a fine-tuned version of HuggingFaceH4/mistral-7b-sft-beta on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5029
• Rewards/chosen: -1.1300
• Rewards/rejected: -1.9954
• Rewards/accuracies: 0.75
• Rewards/margins: 0.8654
• Logps/rejected: -427.3289
• Logps/chosen: -335.3676
• Logits/rejected: -1.9689
• Logits/chosen: -2.0651

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: 2
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• 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: linear
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5625	0.2060	100	0.5413	-1.0460	-1.5393	0.6875	0.4932	-381.7129	-326.9707	-2.4726	-2.5375
0.5145	0.4119	200	0.5226	-1.0460	-1.7939	0.7656	0.7479	-407.1714	-326.9680	-2.1622	-2.2436
0.4948	0.6179	300	0.5165	-0.9692	-1.7757	0.75	0.8064	-405.3522	-319.2909	-2.0477	-2.1379
0.4931	0.8239	400	0.5051	-1.0779	-1.9119	0.75	0.8340	-418.9738	-330.1579	-2.0159	-2.1084

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.20.0