This model is a fine-tuned version of tsavage68/UTI_M2_1000steps_1e7rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6867
Rewards/chosen: -0.0017
Rewards/rejected: -0.0154
Rewards/accuracies: 0.1700
Rewards/margins: 0.0138
Logps/rejected: -9.4048
Logps/chosen: -4.5458
Logits/rejected: -2.7057
Logits/chosen: -2.7050

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-08
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: 275

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6931	0.3333	25	0.6941	0.0016	0.0029	0.0700	-0.0013	-9.3682	-4.5393	-2.7069	-2.7061
0.6849	0.6667	50	0.6964	-0.0083	-0.0028	0.1100	-0.0055	-9.3796	-4.5591	-2.7057	-2.7049
0.6934	1.0	75	0.6896	-0.0050	-0.0129	0.1300	0.0079	-9.3998	-4.5524	-2.7063	-2.7056
0.6902	1.3333	100	0.6901	-0.0010	-0.0078	0.1400	0.0068	-9.3896	-4.5445	-2.7066	-2.7058
0.6942	1.6667	125	0.6876	0.0031	-0.0090	0.1400	0.0121	-9.3920	-4.5362	-2.7061	-2.7053
0.6823	2.0	150	0.6875	0.0028	-0.0094	0.1500	0.0122	-9.3928	-4.5369	-2.7062	-2.7055
0.6846	2.3333	175	0.6803	0.0047	-0.0227	0.1700	0.0273	-9.4193	-4.5331	-2.7064	-2.7057
0.6766	2.6667	200	0.6874	-0.0018	-0.0138	0.1600	0.0120	-9.4015	-4.5461	-2.7058	-2.7050
0.6896	3.0	225	0.6873	0.0001	-0.0126	0.1500	0.0127	-9.3992	-4.5423	-2.7057	-2.7050
0.6895	3.3333	250	0.6867	-0.0017	-0.0154	0.1700	0.0138	-9.4048	-4.5458	-2.7057	-2.7050
0.687	3.6667	275	0.6867	-0.0017	-0.0154	0.1700	0.0138	-9.4048	-4.5458	-2.7057	-2.7050

Framework versions

Transformers 4.41.2
Pytorch 2.0.0+cu117
Datasets 2.19.2
Tokenizers 0.19.1

tsavage68/UTI2_M2_275steps_1e8rate_05beta_CSFTDPO

UTI2_M2_275steps_1e8rate_05beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
UTI2_M2_275steps_1e8rate_05beta_CSFTDPO