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: 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
0.7019	0.2	25	0.6786	-0.4349	-0.4775	0.5800	0.0427	-20.1465	-19.9838	-1.0698	-1.0685
0.6126	0.4	50	0.3723	2.2282	1.0718	0.8800	1.1564	-14.9821	-11.1070	-1.0337	-1.0340
0.6918	0.6	75	0.0759	3.7913	-5.7738	0.9900	9.5651	-37.8007	-5.8967	-1.0561	-1.0499
0.4405	0.8	100	0.1383	3.7679	-5.8296	0.9800	9.5975	-37.9869	-5.9747	-1.0324	-1.0129
0.0	1.0	125	0.0000	4.4890	-8.1495	1.0	12.6386	-45.7198	-3.5708	-1.0484	-1.0252
0.3231	1.2	150	0.0000	4.0358	-7.2360	1.0	11.2718	-42.6746	-5.0815	-1.0250	-1.0111
0.0	1.4	175	0.0000	4.0896	-7.5543	1.0	11.6440	-43.7358	-4.9021	-1.0259	-1.0117
0.0	1.6	200	0.0000	4.1143	-7.6011	1.0	11.7154	-43.8917	-4.8197	-1.0232	-1.0079
0.0	1.8	225	0.0000	4.1494	-7.8210	1.0	11.9704	-44.6247	-4.7029	-1.0248	-1.0090
0.0	2.0	250	0.0000	4.1728	-7.9893	1.0	12.1621	-45.1857	-4.6248	-1.0246	-1.0084
0.0	2.2	275	0.0000	4.1928	-8.1237	1.0	12.3165	-45.6337	-4.5581	-1.0264	-1.0096
0.0	2.4	300	0.0000	4.2148	-8.2668	1.0	12.4816	-46.1106	-4.4848	-1.0260	-1.0090
0.0	2.6	325	0.0000	4.2269	-8.3679	1.0	12.5948	-46.4476	-4.4445	-1.0271	-1.0096
0.0	2.8	350	0.0000	4.2398	-8.4655	1.0	12.7052	-46.7729	-4.4017	-1.0280	-1.0104
0.0	3.0	375	0.0000	4.2594	-8.5477	1.0	12.8071	-47.0470	-4.3361	-1.0290	-1.0111
0.0	3.2	400	0.0000	4.2724	-8.6275	1.0	12.8999	-47.3132	-4.2929	-1.0279	-1.0097
0.0	3.4	425	0.0000	4.2786	-8.7058	1.0	12.9843	-47.5739	-4.2724	-1.0293	-1.0109
0.0	3.6	450	0.0000	4.2937	-8.7547	1.0	13.0484	-47.7369	-4.2217	-1.0293	-1.0109
0.0	3.8	475	0.0000	4.2991	-8.8078	1.0	13.1069	-47.9139	-4.2038	-1.0292	-1.0105
0.0	4.0	500	0.0000	4.3050	-8.8538	1.0	13.1587	-48.0672	-4.1843	-1.0295	-1.0106
0.0	4.2	525	0.0000	4.3101	-8.9007	1.0	13.2108	-48.2237	-4.1672	-1.0296	-1.0107
0.0	4.4	550	0.0000	4.3169	-8.9396	1.0	13.2565	-48.3533	-4.1444	-1.0297	-1.0107
0.0	4.6	575	0.0000	4.3210	-8.9802	1.0	13.3012	-48.4887	-4.1310	-1.0302	-1.0112
0.0	4.8	600	0.0000	4.3255	-9.0068	1.0	13.3324	-48.5775	-4.1158	-1.0313	-1.0122
0.0	5.0	625	0.0000	4.3270	-9.0307	1.0	13.3577	-48.6569	-4.1109	-1.0311	-1.0119
0.0	5.2	650	0.0000	4.3299	-9.0577	1.0	13.3876	-48.7470	-4.1011	-1.0302	-1.0109
0.0	5.4	675	0.0000	4.3329	-9.0561	1.0	13.3890	-48.7417	-4.0911	-1.0311	-1.0119
0.0	5.6	700	0.0000	4.3363	-9.0811	1.0	13.4174	-48.8251	-4.0798	-1.0301	-1.0108
0.0	5.8	725	0.0000	4.3362	-9.0984	1.0	13.4345	-48.8826	-4.0804	-1.0306	-1.0113
0.0	6.0	750	0.0000	4.3386	-9.0976	1.0	13.4362	-48.8800	-4.0721	-1.0307	-1.0112
0.0	6.2	775	0.0000	4.3370	-9.1127	1.0	13.4497	-48.9305	-4.0777	-1.0318	-1.0123
0.0	6.4	800	0.0000	4.3378	-9.1143	1.0	13.4522	-48.9358	-4.0749	-1.0308	-1.0114
0.0	6.6	825	0.0000	4.3371	-9.1165	1.0	13.4536	-48.9429	-4.0771	-1.0310	-1.0116
0.0	6.8	850	0.0000	4.3373	-9.1236	1.0	13.4609	-48.9667	-4.0765	-1.0307	-1.0113
0.0	7.0	875	0.0000	4.3377	-9.1190	1.0	13.4566	-48.9512	-4.0753	-1.0310	-1.0116
0.0	7.2	900	0.0000	4.3356	-9.1212	1.0	13.4568	-48.9586	-4.0821	-1.0310	-1.0117
0.0	7.4	925	0.0000	4.3352	-9.1260	1.0	13.4613	-48.9748	-4.0834	-1.0304	-1.0110
0.0	7.6	950	0.0000	4.3358	-9.1251	1.0	13.4609	-48.9718	-4.0815	-1.0303	-1.0110
0.0	7.8	975	0.0000	4.3354	-9.1238	1.0	13.4592	-48.9674	-4.0829	-1.0304	-1.0110
0.0	8.0	1000	0.0000	4.3354	-9.1238	1.0	13.4592	-48.9674	-4.0829	-1.0304	-1.0110

Framework versions

Transformers 4.40.2
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

tsavage68/Transaminitis_L3_1000steps_1e6rate_03beta_CSFTDPO

Transaminitis_L3_1000steps_1e6rate_03beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
Transaminitis_L3_1000steps_1e6rate_03beta_CSFTDPO