walamakan-t5-base

This model is a fine-tuned version of Karzan/walamakan-t5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3291
• Bleu: 0.0
• Gen Len: 19.0

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	0.96	21	1.3668	0.0	19.0
No log	1.97	43	1.3613	0.0	19.0
No log	2.98	65	1.3530	0.0	19.0
No log	3.99	87	1.3503	0.0	19.0
No log	5.0	109	1.3476	0.0	19.0
No log	5.96	130	1.3429	0.0	19.0
No log	6.97	152	1.3415	0.0	19.0
No log	7.98	174	1.3408	0.0	19.0
No log	8.99	196	1.3382	0.0	19.0
No log	9.99	218	1.3387	0.0	19.0
No log	10.96	239	1.3301	0.0	19.0
No log	11.97	261	1.3344	0.0	19.0
No log	12.97	283	1.3323	0.0	19.0
No log	13.98	305	1.3312	0.0	19.0
No log	14.99	327	1.3329	0.0	19.0
No log	16.0	349	1.3323	0.0	19.0
No log	16.96	370	1.3289	0.0	19.0
No log	17.97	392	1.3291	0.0	19.0
No log	18.98	414	1.3289	0.0	19.0
No log	19.26	420	1.3291	0.0	19.0

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3