This model is a fine-tuned version of tsavage68/chat_600STEPS_1e8rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6935
Rewards/chosen: -0.0079
Rewards/rejected: -0.0079
Rewards/accuracies: 0.4286
Rewards/margins: -0.0000
Logps/rejected: -18.8283
Logps/chosen: -16.7810
Logits/rejected: -0.5983
Logits/chosen: -0.5982

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-07
train_batch_size: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 650

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6925	0.0977	50	0.6943	0.0013	0.0029	0.3846	-0.0016	-18.7922	-16.7503	-0.5979	-0.5978
0.6919	0.1953	100	0.6932	-0.0001	-0.0007	0.4110	0.0005	-18.8042	-16.7551	-0.5986	-0.5985
0.6907	0.2930	150	0.6939	-0.0044	-0.0036	0.4198	-0.0008	-18.8141	-16.7693	-0.5983	-0.5982
0.6943	0.3906	200	0.6931	-0.0045	-0.0052	0.4198	0.0007	-18.8195	-16.7697	-0.5976	-0.5975
0.6956	0.4883	250	0.6926	-0.0038	-0.0056	0.4396	0.0017	-18.8205	-16.7673	-0.5985	-0.5984
0.6893	0.5859	300	0.6921	-0.0055	-0.0082	0.4022	0.0027	-18.8295	-16.7730	-0.5980	-0.5979
0.6886	0.6836	350	0.6908	-0.0050	-0.0105	0.4484	0.0054	-18.8369	-16.7714	-0.5979	-0.5978
0.6909	0.7812	400	0.6908	-0.0036	-0.0092	0.4198	0.0056	-18.8326	-16.7665	-0.5984	-0.5983
0.6882	0.8789	450	0.6927	-0.0075	-0.0091	0.4264	0.0016	-18.8322	-16.7795	-0.5983	-0.5982
0.6907	0.9766	500	0.6911	-0.0053	-0.0101	0.4484	0.0048	-18.8357	-16.7724	-0.5984	-0.5983
0.6897	1.0742	550	0.6932	-0.0076	-0.0082	0.4110	0.0005	-18.8293	-16.7801	-0.5983	-0.5982
0.6826	1.1719	600	0.6916	-0.0047	-0.0085	0.4593	0.0038	-18.8302	-16.7702	-0.5981	-0.5980
0.6857	1.2695	650	0.6935	-0.0079	-0.0079	0.4286	-0.0000	-18.8283	-16.7810	-0.5983	-0.5982

Framework versions

Transformers 4.40.1
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

tsavage68/chat_650_STEPS_03beta_1e7rate_CDPOSFT

chat_650_STEPS_03beta_1e7rate_CDPOSFT

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
chat_650_STEPS_03beta_1e7rate_CDPOSFT