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: 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: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6831	0.2004	50	0.6816	0.0015	-0.0238	0.1300	0.0253	-15.3431	-9.3779	-1.0962	-1.0977
0.6795	0.4008	100	0.6463	0.0093	-0.1112	0.1400	0.1205	-15.6344	-9.3518	-1.0932	-1.0948
0.6329	0.6012	150	0.6076	0.0323	-0.3453	0.1400	0.3776	-16.4149	-9.2751	-1.0926	-1.0943
0.6091	0.8016	200	0.5997	0.0442	-0.5668	0.1400	0.6110	-17.1532	-9.2355	-1.0949	-1.0965
0.6241	1.0020	250	0.5974	0.0514	-0.7694	0.1400	0.8208	-17.8283	-9.2113	-1.0983	-1.0999
0.6239	1.2024	300	0.5969	0.0644	-0.8984	0.1400	0.9628	-18.2584	-9.1680	-1.1014	-1.1028
0.624	1.4028	350	0.5965	0.0676	-0.9908	0.1400	1.0585	-18.5665	-9.1573	-1.1032	-1.1046
0.5728	1.6032	400	0.5965	0.0722	-1.0529	0.1400	1.1250	-18.7733	-9.1423	-1.1052	-1.1066
0.5893	1.8036	450	0.5964	0.0748	-1.0956	0.1400	1.1704	-18.9158	-9.1336	-1.1062	-1.1075
0.5719	2.0040	500	0.5964	0.0693	-1.1155	0.1400	1.1848	-18.9820	-9.1518	-1.1066	-1.1079
0.5719	2.2044	550	0.5964	0.0760	-1.1221	0.1400	1.1981	-19.0042	-9.1295	-1.1069	-1.1082
0.5546	2.4048	600	0.5964	0.0686	-1.1465	0.1400	1.2151	-19.0856	-9.1542	-1.1071	-1.1084
0.52	2.6052	650	0.5964	0.0707	-1.1510	0.1400	1.2217	-19.1005	-9.1471	-1.1066	-1.1079
0.6243	2.8056	700	0.5963	0.0745	-1.1541	0.1400	1.2286	-19.1107	-9.1345	-1.1075	-1.1088
0.6065	3.0060	750	0.5963	0.0758	-1.1510	0.1400	1.2268	-19.1006	-9.1301	-1.1071	-1.1084
0.6412	3.2064	800	0.5964	0.0704	-1.1555	0.1400	1.2259	-19.1153	-9.1480	-1.1070	-1.1083
0.6585	3.4068	850	0.5963	0.0726	-1.1522	0.1400	1.2248	-19.1045	-9.1408	-1.1073	-1.1086
0.6238	3.6072	900	0.5963	0.0735	-1.1585	0.1400	1.2320	-19.1256	-9.1378	-1.1071	-1.1084
0.5372	3.8076	950	0.5964	0.0711	-1.1551	0.1400	1.2262	-19.1142	-9.1459	-1.1071	-1.1083
0.6239	4.0080	1000	0.5964	0.0711	-1.1551	0.1400	1.2262	-19.1142	-9.1459	-1.1071	-1.1083

Framework versions

Transformers 4.41.2
Pytorch 2.0.0+cu117
Datasets 2.20.0
Tokenizers 0.19.1

tsavage68/Summary_L3_1000steps_1e7rate_03beta_CSFTDPO

Summary_L3_1000steps_1e7rate_03beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
Summary_L3_1000steps_1e7rate_03beta_CSFTDPO