bangla_idiom_generation_v2

This model is a fine-tuned version of mHossain/bangla-para-v3-500000 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 4.7655
• Rouge1: 0.0
• Rouge2: 0.0
• Rougel: 0.0
• Rougelsum: 0.0
• Gen Len: 10.4292

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	226	7.4650	0.0	0.0	0.0	0.0	11.1195
No log	2.0	452	7.1989	0.0	0.0	0.0	0.0	10.677
7.7649	3.0	678	6.8146	0.0	0.0	0.0	0.0	9.8938
7.7649	4.0	904	6.3650	0.0	0.0	0.0	0.0	9.3805
7.0991	5.0	1130	5.8781	0.0	0.0	0.0	0.0	9.2478
7.0991	6.0	1356	5.4460	0.0	0.0	0.0	0.0	9.0487
6.1699	7.0	1582	5.1824	0.0	0.0	0.0	0.0	9.3363
6.1699	8.0	1808	4.9930	0.0	0.0	0.0	0.0	10.2478
5.5395	9.0	2034	4.8596	0.0	0.0	0.0	0.0	10.5575
5.5395	10.0	2260	4.7655	0.0	0.0	0.0	0.0	10.4292

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3