This model is a fine-tuned version of tsavage68/IE_M2_1000steps_1e7rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6608
Rewards/chosen: 0.0105
Rewards/rejected: -0.0599
Rewards/accuracies: 0.3850
Rewards/margins: 0.0703
Logps/rejected: -41.1416
Logps/chosen: -42.1846
Logits/rejected: -2.9156
Logits/chosen: -2.8543

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: 350

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6945	0.4	50	0.6933	0.0121	0.0098	0.2450	0.0023	-41.0022	-42.1813	-2.9159	-2.8545
0.6936	0.8	100	0.6888	0.0052	-0.0069	0.2150	0.0121	-41.0356	-42.1952	-2.9158	-2.8545
0.6822	1.2	150	0.6646	0.0021	-0.0599	0.3650	0.0621	-41.1417	-42.2012	-2.9157	-2.8544
0.6637	1.6	200	0.6652	0.0023	-0.0586	0.3600	0.0609	-41.1390	-42.2010	-2.9157	-2.8544
0.6647	2.0	250	0.6601	0.0043	-0.0670	0.3900	0.0713	-41.1557	-42.1968	-2.9157	-2.8544
0.6697	2.4	300	0.6624	0.0067	-0.0606	0.3800	0.0673	-41.1430	-42.1922	-2.9156	-2.8543
0.6579	2.8	350	0.6608	0.0105	-0.0599	0.3850	0.0703	-41.1416	-42.1846	-2.9156	-2.8543

Framework versions

Transformers 4.44.2
Pytorch 2.0.0+cu117
Datasets 3.0.0
Tokenizers 0.19.1

tsavage68/IE_M2_350steps_1e8rate_05beta_cSFTDPO

IE_M2_350steps_1e8rate_05beta_cSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
IE_M2_350steps_1e8rate_05beta_cSFTDPO