This model is a fine-tuned version of tsavage68/UTI_L3_1000steps_1e5rate_SFT on an unknown dataset.
It achieves the following results on the evaluation set:
eval_loss: 0.0077
eval_runtime: 14.9161
eval_samples_per_second: 6.704
eval_steps_per_second: 6.704
eval_rewards/chosen: 2.2023
eval_rewards/rejected: -11.8065
eval_rewards/accuracies: 0.9900
eval_rewards/margins: 14.0088
eval_logps/rejected: -102.5496
eval_logps/chosen: -25.1379
eval_logits/rejected: -1.3939
eval_logits/chosen: -1.3632
epoch: 13.3333
step: 1000
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
