tinyllama-chat-mine-dpo

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6240
• Rewards/chosen: -0.7007
• Rewards/rejected: -0.9684
• Rewards/accuracies: 0.6825
• Rewards/margins: 0.2677
• Logps/rejected: -395.3808
• Logps/chosen: -412.6868
• Logits/rejected: -2.7107
• Logits/chosen: -2.7399

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6685	0.2093	100	0.6694	-0.1299	-0.1943	0.6528	0.0644	-317.9682	-355.6027	-2.9255	-2.9507
0.642	0.4186	200	0.6407	-0.4273	-0.6175	0.6726	0.1902	-360.2873	-385.3470	-2.7926	-2.8208
0.6285	0.6279	300	0.6331	-0.4723	-0.6951	0.6647	0.2228	-368.0482	-389.8438	-2.7731	-2.8012
0.6222	0.8373	400	0.6240	-0.7007	-0.9684	0.6825	0.2677	-395.3808	-412.6868	-2.7107	-2.7399

Framework versions

• Transformers 4.43.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1