This model is a fine-tuned version of Minbyul/selfbiorag-7b-wo-healthsearch_qa-sft on the HuggingFaceH4/ultrafeedback_binarized dataset.
It achieves the following results on the evaluation set:
Loss: 0.4465
Rewards/chosen: -0.5929
Rewards/rejected: -1.6772
Rewards/accuracies: 0.7846
Rewards/margins: 1.0843
Logps/rejected: -1480.8429
Logps/chosen: -984.8102
Logits/rejected: -3.4642
Logits/chosen: -2.6475
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: 5e-06
train_batch_size: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 2
total_train_batch_size: 64
total_eval_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
