This model is a fine-tuned version of tsavage68/UTI_L3_1000steps_1e5rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0100
Rewards/chosen: 1.5940
Rewards/rejected: -8.1393
Rewards/accuracies: 0.9900
Rewards/margins: 9.7333
Logps/rejected: -90.3257
Logps/chosen: -27.1656
Logits/rejected: -1.3519
Logits/chosen: -1.3283

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: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
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: 75

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.593	0.3333	25	0.3798	0.1560	-0.7116	0.9800	0.8675	-65.5666	-31.9591	-1.3242	-1.3092
0.0537	0.6667	50	0.0457	0.8318	-4.9451	0.9900	5.7769	-79.6783	-29.7064	-1.3335	-1.3154
0.0012	1.0	75	0.0100	1.5940	-8.1393	0.9900	9.7333	-90.3257	-27.1656	-1.3519	-1.3283

Framework versions

Transformers 4.41.2
Pytorch 2.0.0+cu117
Datasets 2.19.2
Tokenizers 0.19.1

tsavage68/UTI_L3_75steps_1e6rate_03beta_CSFTDPO

UTI_L3_75steps_1e6rate_03beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
UTI_L3_75steps_1e6rate_03beta_CSFTDPO