File size: 1,914 Bytes
14f69bc |
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 |
---
license: apache-2.0
base_model: lidiya/bart-large-xsum-samsum
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: tech-dialogue-summarization-3
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. -->
# tech-dialogue-summarization-3
This model is a fine-tuned version of [lidiya/bart-large-xsum-samsum](https://huggingface.co./lidiya/bart-large-xsum-samsum) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3468
- Rouge1: 52.6316
- Rouge2: 32.4324
- Rougel: 47.3684
- Rougelsum: 47.3684
- Gen Len: 38.0
## 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
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- 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 | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
| No log | 1.0 | 1 | 1.4146 | 52.6316 | 32.4324 | 47.3684 | 47.3684 | 38.0 |
| No log | 2.0 | 2 | 1.3652 | 52.6316 | 32.4324 | 47.3684 | 47.3684 | 38.0 |
| No log | 3.0 | 3 | 1.3468 | 52.6316 | 32.4324 | 47.3684 | 47.3684 | 38.0 |
### Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1
|