---
library_name: transformers
base_model: CausalNLP/gpt2-hf_multilingual-20
tags:
- generated_from_trainer
model-index:
- name: gpt2-multilingual-20-arabic-repair_3epochs_lr1e-4_bf16
  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. -->

# gpt2-multilingual-20-arabic-repair_3epochs_lr1e-4_bf16

This model is a fine-tuned version of [CausalNLP/gpt2-hf_multilingual-20](https://huggingface.co/CausalNLP/gpt2-hf_multilingual-20) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2206

## 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: 0.0001
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Use adamw_torch_fused with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3

### Training results

| Training Loss | Epoch  | Step  | Validation Loss |
|:-------------:|:------:|:-----:|:---------------:|
| 3.295         | 0.0799 | 500   | 3.3488          |
| 3.2752        | 0.1597 | 1000  | 3.3363          |
| 3.2844        | 0.2396 | 1500  | 3.3416          |
| 3.3059        | 0.3194 | 2000  | 3.3491          |
| 3.335         | 0.3993 | 2500  | 3.3387          |
| 3.2845        | 0.4791 | 3000  | 3.3301          |
| 3.2458        | 0.5590 | 3500  | 3.3216          |
| 3.2689        | 0.6388 | 4000  | 3.3129          |
| 3.2737        | 0.7187 | 4500  | 3.3069          |
| 3.2604        | 0.7985 | 5000  | 3.3000          |
| 3.2255        | 0.8784 | 5500  | 3.2936          |
| 3.239         | 0.9582 | 6000  | 3.2882          |
| 3.2594        | 1.0380 | 6500  | 3.2832          |
| 3.2123        | 1.1179 | 7000  | 3.2780          |
| 3.2347        | 1.1977 | 7500  | 3.2726          |
| 3.2197        | 1.2776 | 8000  | 3.2671          |
| 3.2028        | 1.3574 | 8500  | 3.2615          |
| 3.195         | 1.4373 | 9000  | 3.2557          |
| 3.2049        | 1.5171 | 9500  | 3.2510          |
| 3.1649        | 1.5970 | 10000 | 3.2458          |
| 3.1984        | 1.6768 | 10500 | 3.2409          |
| 3.2787        | 1.7567 | 11000 | 3.2374          |
| 3.1966        | 1.8365 | 11500 | 3.2334          |
| 3.166         | 1.9164 | 12000 | 3.2305          |
| 3.2029        | 1.9962 | 12500 | 3.2272          |
| 3.1629        | 2.0760 | 13000 | 3.2268          |
| 3.1115        | 2.1559 | 13500 | 3.2253          |
| 3.184         | 2.2357 | 14000 | 3.2239          |
| 3.1391        | 2.3156 | 14500 | 3.2229          |
| 3.1341        | 2.3954 | 15000 | 3.2220          |
| 3.1623        | 2.4753 | 15500 | 3.2214          |
| 3.1565        | 2.5551 | 16000 | 3.2211          |
| 3.1129        | 2.6350 | 16500 | 3.2208          |
| 3.2043        | 2.7148 | 17000 | 3.2207          |
| 3.1044        | 2.7947 | 17500 | 3.2206          |
| 3.1922        | 2.8746 | 18000 | 3.2206          |
| 3.1229        | 2.9544 | 18500 | 3.2206          |


### Framework versions

- Transformers 4.57.4
- Pytorch 2.9.1+cu128
- Datasets 4.4.2
- Tokenizers 0.22.2