File size: 1,952 Bytes
2fcc175 |
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 |
---
license: apache-2.0
base_model: t5-base
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: pargraphs_titles
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. -->
# pargraphs_titles
This model is a fine-tuned version of [t5-base](https://huggingface.co/t5-base) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2408
- Rouge1: 68.8334
- Rouge2: 59.1001
- Rougel: 67.5283
- Rougelsum: 67.6312
- Gen Len: 13.7977
## 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: 4e-05
- train_batch_size: 4
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
| No log | 1.0 | 455 | 0.4086 | 56.8689 | 47.8503 | 56.3172 | 56.1544 | 13.7393 |
| 0.2169 | 2.0 | 910 | 0.1848 | 66.7857 | 58.5467 | 65.8133 | 65.7037 | 13.8604 |
| 0.1629 | 3.0 | 1365 | 0.1768 | 67.5534 | 59.4926 | 66.5736 | 66.5271 | 13.7222 |
| 0.1347 | 4.0 | 1820 | 0.2881 | 67.7423 | 58.9397 | 66.4095 | 66.5328 | 13.8362 |
| 0.1277 | 5.0 | 2275 | 0.2408 | 68.8334 | 59.1001 | 67.5283 | 67.6312 | 13.7977 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Tokenizers 0.15.0
|