Summary_L3_200steps_1e6rate_03beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Summary_L3_1000steps_1e7rate_SFT2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5961
• Rewards/chosen: 0.1275
• Rewards/rejected: -2.3694
• Rewards/accuracies: 0.1400
• Rewards/margins: 2.4969
• Logps/rejected: -20.0026
• Logps/chosen: -9.1277
• Logits/rejected: -1.1104
• Logits/chosen: -1.1118

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: 1e-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 200

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.555	0.2004	50	0.5962	0.0976	-1.3577	0.1400	1.4553	-17.9791	-9.1876	-1.0985	-1.1002
0.6585	0.4008	100	0.5962	0.1094	-1.5231	0.1400	1.6326	-18.3100	-9.1639	-1.1003	-1.1019
0.6238	0.6012	150	0.5961	0.1393	-2.3520	0.1400	2.4913	-19.9677	-9.1042	-1.1096	-1.1111
0.6065	0.8016	200	0.5961	0.1275	-2.3694	0.1400	2.4969	-20.0026	-9.1277	-1.1104	-1.1118

Framework versions

• Transformers 4.41.2
• Pytorch 2.0.0+cu117
• Datasets 2.20.0
• Tokenizers 0.19.1