File size: 4,197 Bytes
50ea912 3e541c2 50ea912 fd88a9e 50ea912 |
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 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- winograd_wsc
metrics:
- rouge
model-index:
- name: flan-t5-large-coref
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: winograd_wsc
type: winograd_wsc
config: wsc285
split: test
args: wsc285
metrics:
- name: Rouge1
type: rouge
value: 0.9495
widget:
- text: "Sam has a Parker pen. He loves writing with it."
example_title: "Example 1"
- text: "Coronavirus quickly spread worldwide in 2020. The virus mostly affects elderly people. They can easily catch it."
example_title: "Example 2"
- text: "First, the manager evaluates the candidates. Afterwards, he notifies the candidates regarding the evaluation."
example_title: "Example 3"
---
<!-- 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. -->
# flan-t5-large-coref
This model is a fine-tuned version of [google/flan-t5-large](https://huggingface.co/google/flan-t5-large) on the winograd_wsc dataset.
The model was trained on the task of coreference resolution.
It achieves the following results on the evaluation set:
- Loss: 0.2404
- Rouge1: 0.9495
- Rouge2: 0.9107
- Rougel: 0.9494
- Rougelsum: 0.9494
- Gen Len: 23.4828
## 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: 16
- eval_batch_size: 16
- seed: 42
- 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 | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:-------:|
| 1.0169 | 1.0 | 16 | 0.6742 | 0.7918 | 0.6875 | 0.7836 | 0.7847 | 18.2414 |
| 0.6275 | 2.0 | 32 | 0.5093 | 0.8776 | 0.7947 | 0.8734 | 0.8732 | 21.5517 |
| 0.596 | 3.0 | 48 | 0.4246 | 0.9104 | 0.8486 | 0.9085 | 0.9091 | 22.5172 |
| 0.743 | 4.0 | 64 | 0.3632 | 0.9247 | 0.8661 | 0.9235 | 0.9231 | 22.8621 |
| 0.5007 | 5.0 | 80 | 0.3301 | 0.9353 | 0.8845 | 0.9357 | 0.9353 | 22.8621 |
| 0.2567 | 6.0 | 96 | 0.3093 | 0.9388 | 0.8962 | 0.9392 | 0.9388 | 22.9655 |
| 0.4146 | 7.0 | 112 | 0.2978 | 0.9449 | 0.907 | 0.9455 | 0.9458 | 23.1034 |
| 0.1991 | 8.0 | 128 | 0.2853 | 0.9454 | 0.9064 | 0.946 | 0.9462 | 23.069 |
| 0.1786 | 9.0 | 144 | 0.2794 | 0.9475 | 0.9097 | 0.9475 | 0.9477 | 23.069 |
| 0.3559 | 10.0 | 160 | 0.2701 | 0.9424 | 0.9013 | 0.9428 | 0.9426 | 23.0345 |
| 0.2059 | 11.0 | 176 | 0.2636 | 0.9472 | 0.9069 | 0.9472 | 0.9472 | 23.0345 |
| 0.199 | 12.0 | 192 | 0.2592 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.1634 | 13.0 | 208 | 0.2553 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.2006 | 14.0 | 224 | 0.2518 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.1419 | 15.0 | 240 | 0.2487 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.2089 | 16.0 | 256 | 0.2456 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.1007 | 17.0 | 272 | 0.2431 | 0.9523 | 0.9141 | 0.9521 | 0.9524 | 23.4483 |
| 0.1598 | 18.0 | 288 | 0.2415 | 0.9495 | 0.9107 | 0.9494 | 0.9494 | 23.4828 |
| 0.3088 | 19.0 | 304 | 0.2407 | 0.9495 | 0.9107 | 0.9494 | 0.9494 | 23.4828 |
| 0.2003 | 20.0 | 320 | 0.2404 | 0.9495 | 0.9107 | 0.9494 | 0.9494 | 23.4828 |
### Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.7.1
- Tokenizers 0.13.2
|