File size: 2,059 Bytes
b8010e8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
---
language:
- ko
- ja
base_model: facebook/mbart-large-50-many-to-many-mmt
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: koja_reverse_13p_run1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# koja_reverse_13p_run1
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3532
- Bleu: 5.7574
- Gen Len: 20.4109
## 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: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- num_epochs: 10
### Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:------:|:-------:|
| 1.9693 | 0.42 | 800 | 1.5445 | 4.3096 | 20.2019 |
| 1.4914 | 0.83 | 1600 | 1.4233 | 4.307 | 20.1204 |
| 1.4272 | 1.25 | 2400 | 1.3761 | 4.8111 | 22.8555 |
| 1.1571 | 1.66 | 3200 | 1.3532 | 5.7574 | 21.151 |
| 1.0833 | 2.08 | 4000 | 1.3449 | 5.652 | 21.2158 |
| 0.9394 | 2.49 | 4800 | 1.3905 | 5.45 | 19.9005 |
| 0.8633 | 2.91 | 5600 | 1.3719 | 4.9973 | 19.7844 |
### Framework versions
- Transformers 4.34.1
- Pytorch 2.1.0+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1
|