zephyr-7b-dpo-full-ultrabin-3-avg-logprob-lr-same

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6219
• Rewards/chosen: -0.2287
• Rewards/rejected: -0.4214
• Rewards/accuracies: 0.7188
• Rewards/margins: 0.1928
• Logps/rejected: -43.2637
• Logps/chosen: -23.8237
• Logits/rejected: 2.7050
• Logits/chosen: 2.7595

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: 55
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• 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
0.6803	0.2092	100	0.6741	-0.1830	-0.2238	0.6055	0.0408	-23.4969	-19.2551	5.1619	4.6722
0.6333	0.4184	200	0.6318	-0.2607	-0.4263	0.7070	0.1656	-43.7487	-27.0254	2.0618	2.2059
0.6202	0.6276	300	0.6256	-0.1960	-0.3896	0.7188	0.1936	-40.0823	-20.5594	3.0356	2.9998
0.6195	0.8368	400	0.6219	-0.2287	-0.4214	0.7188	0.1928	-43.2637	-23.8237	2.7050	2.7595

Framework versions

• Transformers 4.44.0.dev0
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1