aiden200
/

anon

Video-Text-to-Text

Generated from Trainer

Model card Files Files and versions

aiden200 commited on Apr 24

Commit

6451b9d

·

verified ·

1 Parent(s): 56278bd

Model save

Files changed (2) hide show

README.md +2 -2
train.log +1 -0

README.md CHANGED Viewed

@@ -39,8 +39,8 @@ The following hyperparameters were used during training:
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 4
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 32
 - total_eval_batch_size: 4
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine

 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 4
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 8
 - total_eval_batch_size: 4
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine

train.log CHANGED Viewed

@@ -16634,3 +16634,4 @@ tensor(-0.0011, device='cuda:1', grad_fn=<AddBackward0>) tensor(-0.0011, device=
 Finished TrainingFinished Training
 Finished Training

 Finished TrainingFinished Training
 Finished Training
+Finished Training