This model is a fine-tuned version of llava-hf/llava-v1.6-mistral-7b-hf on the AA_preference_Cherry_0_80 dataset. It achieves the following results on the evaluation set:

Loss: 0.4646
Rewards/chosen: 1.3854
Rewards/rejected: -2.0318
Rewards/accuracies: 0.8495
Rewards/margins: 3.4173
Logps/rejected: -243.1421
Logps/chosen: -276.4326
Logits/rejected: -2.0370
Logits/chosen: -2.0845

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: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 4
total_train_batch_size: 256
total_eval_batch_size: 64
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 10
num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5761	0.9259	100	0.4648	1.1012	-1.0931	0.8061	2.1944	-233.7547	-279.2745	-2.1461	-2.1677
0.2547	1.8519	200	0.4681	1.5677	-1.4567	0.8367	3.0244	-237.3907	-274.6099	-2.0097	-2.0551
0.1018	2.7778	300	0.4645	1.3964	-2.0290	0.8495	3.4254	-243.1136	-276.3229	-2.0408	-2.0880

Framework versions

Transformers 4.45.2
Pytorch 2.4.0+cu121
Datasets 2.21.0
Tokenizers 0.20.3

htlou/backup_0202_llamafactory_AA_preference_Cherry_0_80-llava-mistral

AA_preference_Cherry_0_80

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

htlou/
backup_0202_llamafactory_AA_preference_Cherry_0_80-llava-mistral