metadata

license: apache-2.0
tags:
  - generated_from_keras_callback
model-index:
  - name: suarkadipa/GPT-2-finetuned-papers
    results: []
datasets:
  - CShorten/ML-ArXiv-Papers
language:
  - en
base_model:
  - distilbert/distilgpt2

suarkadipa/GPT-2-finetuned-papers

This model is a fine-tuned version of distilgpt2 on an CShorten/ML-ArXiv-Papers dataset. Based on https://python.plainenglish.io/i-fine-tuned-gpt-2-on-100k-scientific-papers-heres-the-result-903f0784fe65 It achieves the following results on the evaluation set:

Train Loss: 2.4225
Validation Loss: 2.2164
Epoch: 0

Model description

More information needed

Intended uses & limitations

How to run in Google Colab

from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer_fromhub = AutoTokenizer.from_pretrained("suarkadipa/GPT-2-finetuned-papers")
model_fromhub = AutoModelForCausalLM.from_pretrained("suarkadipa/GPT-2-finetuned-papers", from_tf=True)

text_generator = pipeline(
    "text-generation",
    model=model_fromhub,
    tokenizer=tokenizer_fromhub,
    framework="tf",
    max_new_tokens=3000
)

// change with your text
test_sentence = "the role of recommender systems"
res=text_generator(test_sentence)[0]["generated_text"].replace("\n", " ")
res

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

optimizer: {'name': 'AdamWeightDecay', 'learning_rate': {'class_name': 'ExponentialDecay', 'config': {'initial_learning_rate': 0.0005, 'decay_steps': 500, 'decay_rate': 0.95, 'staircase': False, 'name': None}}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01}
training_precision: float32

Training results

Train Loss	Validation Loss	Epoch
2.4225	2.2164	0

Framework versions

Transformers 4.28.1
TensorFlow 2.12.0
Datasets 2.11.0
Tokenizers 0.13.3