mlm_final

This model is a fine-tuned version of gpt2 on a custom dataset using the Digital Image Processing textbook (Gonzalez and Woods, 2018). It achieves the following results on the evaluation set, which used the Fundamentals of Digital Image Processing textbook (Solomon and Breckon, 2010):

• Loss: 4.0700
• Perplexity: 58.6

Model description

This model is trained using Masked Language Modelling.

Intended uses & limitations

This model is intended for use within the field of Computer Vision, as is trained using a Computer Vision textbook.

Training and evaluation data

It is trained and validated using computer vision textbooks split into chunks of 512 tokens

Usage

from transformers import pipeline

question = "What is PCA?"
question_answering = pipeline(model='psxjp5/mlm')
output = question_answering(formatted_text)

print(output[0]['generated_text'])

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 9
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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	Perplexity
15.6719	0.99	22	5.3660	214.0
4.3293	1.98	44	4.4748	87.8
3.882	2.97	66	4.2731	71.7
3.7072	3.96	88	4.1473	63.3
3.6499	4.94	110	4.1219	61.7
3.5604	5.93	132	4.0896	59.7
3.5268	6.92	154	4.0700	58.6

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3