metadata
base_model: facebook/mbart-large-50
library_name: peft
license: mit
tags:
- generated_from_trainer
model-index:
- name: mbart-large-50_Nepali_News_Summarization_QLoRA_8bit
results: []
mbart-large-50_Nepali_News_Summarization_QLoRA_8bit
This model is a fine-tuned version of facebook/mbart-large-50 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3724
- Rouge-1 R: 0.3809
- Rouge-1 P: 0.3877
- Rouge-1 F: 0.3745
- Rouge-2 R: 0.2144
- Rouge-2 P: 0.2176
- Rouge-2 F: 0.2093
- Rouge-l R: 0.3702
- Rouge-l P: 0.3766
- Rouge-l F: 0.364
- Gen Len: 14.0747
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: 0.0005
- train_batch_size: 5
- eval_batch_size: 5
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge-1 R | Rouge-1 P | Rouge-1 F | Rouge-2 R | Rouge-2 P | Rouge-2 F | Rouge-l R | Rouge-l P | Rouge-l F | Gen Len |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.5604 | 1.0 | 10191 | 1.5916 | 0.3605 | 0.3694 | 0.3536 | 0.1948 | 0.2008 | 0.19 | 0.3501 | 0.3586 | 0.3433 | 14.7262 |
| 1.5482 | 2.0 | 20382 | 1.3992 | 0.3673 | 0.3879 | 0.3672 | 0.2034 | 0.2149 | 0.202 | 0.3577 | 0.3775 | 0.3575 | 13.7928 |
| 1.2397 | 3.0 | 30573 | 1.3724 | 0.3809 | 0.3877 | 0.3745 | 0.2144 | 0.2176 | 0.2093 | 0.3702 | 0.3766 | 0.364 | 14.0747 |
Framework versions
- PEFT 0.11.1
- Transformers 4.42.3
- Pytorch 2.1.2
- Datasets 2.20.0
- Tokenizers 0.19.1