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: 0.6853
Rewards/chosen: -0.1288
Rewards/rejected: -0.2807
Rewards/accuracies: 0.5143
Rewards/margins: 0.1518
Logps/rejected: -19.3633
Logps/chosen: -17.0123
Logits/rejected: -0.5890
Logits/chosen: -0.5888

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: 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: 400

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6871	0.0977	50	0.6897	0.0517	0.0417	0.4352	0.0100	-18.7185	-16.6512	-0.6010	-0.6009
0.6399	0.1953	100	0.6728	-0.1560	-0.2548	0.5099	0.0989	-19.3116	-17.0666	-0.6090	-0.6089
0.752	0.2930	150	0.6985	-0.1949	-0.2845	0.4505	0.0896	-19.3710	-17.1445	-0.5936	-0.5934
0.713	0.3906	200	0.6945	-0.1538	-0.2727	0.4923	0.1188	-19.3473	-17.0623	-0.5881	-0.5879
0.7476	0.4883	250	0.6974	-0.1319	-0.2605	0.5165	0.1286	-19.3230	-17.0185	-0.5854	-0.5852
0.6906	0.5859	300	0.6883	-0.1320	-0.2782	0.5165	0.1461	-19.3583	-17.0187	-0.5910	-0.5909
0.6808	0.6836	350	0.6861	-0.1290	-0.2784	0.5077	0.1494	-19.3587	-17.0125	-0.5888	-0.5887
0.6476	0.7812	400	0.6853	-0.1288	-0.2807	0.5143	0.1518	-19.3633	-17.0123	-0.5890	-0.5888

Framework versions

Transformers 4.40.1
Pytorch 2.0.0+cu117
Datasets 2.19.0
Tokenizers 0.19.1

tsavage68/chat_400_STEPS_05beta_1e6rate_CDPOSFT

chat_400_STEPS_05beta_1e6rate_CDPOSFT

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
chat_400_STEPS_05beta_1e6rate_CDPOSFT