---
library_name: transformers
tags:
- generated_from_trainer
metrics:
- rouge
model-index:
- name: gpt22gpt2-gpt2-cnn-dailymail-seed42
  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. -->

# gpt22gpt2-gpt2-cnn-dailymail-seed42

This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 2.3527
- Rouge1: 0.2544
- Rouge2: 0.0686
- Rougel: 0.1560
- Rougelsum: 0.2385

## 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: 8
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 3.0

### Training results

| Training Loss | Epoch  | Step  | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
|:-------------:|:------:|:-----:|:---------------:|:------:|:------:|:------:|:---------:|
| 3.2453        | 0.2229 | 2000  | 3.0408          | 0.2031 | 0.0365 | 0.1269 | 0.1906    |
| 3.0421        | 0.4458 | 4000  | 2.8456          | 0.2380 | 0.0513 | 0.1453 | 0.2232    |
| 2.9227        | 0.6687 | 6000  | 2.7288          | 0.2595 | 0.0617 | 0.1558 | 0.2424    |
| 2.8436        | 0.8916 | 8000  | 2.6556          | 0.2584 | 0.0632 | 0.1555 | 0.2416    |
| 2.6961        | 1.1145 | 10000 | 2.5992          | 0.2578 | 0.0642 | 0.1570 | 0.2410    |
| 2.6662        | 1.3374 | 12000 | 2.5513          | 0.2749 | 0.0717 | 0.1642 | 0.2571    |
| 2.6312        | 1.5603 | 14000 | 2.5081          | 0.2530 | 0.0638 | 0.1543 | 0.2366    |
| 2.6058        | 1.7832 | 16000 | 2.4639          | 0.2636 | 0.0717 | 0.1601 | 0.2469    |
| 2.5725        | 2.0061 | 18000 | 2.4292          | 0.2567 | 0.0689 | 0.1560 | 0.2407    |
| 2.4892        | 2.2290 | 20000 | 2.4027          | 0.2707 | 0.0746 | 0.1640 | 0.2531    |
| 2.4647        | 2.4519 | 22000 | 2.3801          | 0.2508 | 0.0664 | 0.1540 | 0.2350    |
| 2.4479        | 2.6748 | 24000 | 2.3620          | 0.2638 | 0.0727 | 0.1608 | 0.2473    |
| 2.4474        | 2.8977 | 26000 | 2.3527          | 0.2544 | 0.0686 | 0.1560 | 0.2385    |


### Framework versions

- Transformers 4.44.2
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1