Batteryscibert Cased

🚀 BatterySciBERT-cased model

The BatterySciBERT-cased model is a pre-trained model on a large corpus of battery research papers. It starts with the SciBERT-cased weights and uses a masked language modeling (MLM) objective. It was introduced in this paper and first released in this repository. This model is case-sensitive, distinguishing between "english" and "English".

✨ Features

• Pretrained on a large battery research paper corpus using MLM objective.
• Case-sensitive, capable of differentiating between different cases of words.
• Learns a bidirectional representation of sentences, useful for downstream tasks.

📦 Installation

No installation steps are provided in the original README.

💻 Usage Examples

Basic Usage

You can use this model directly with a pipeline for masked language modeling:

>>> from transformers import pipeline
>>> unmasker = pipeline('fill-mask', model='batterydata/batteryscibert-cased')
>>> unmasker("Hello I'm a <mask> model.")

Advanced Usage

Here is how to use this model to get the features of a given text in PyTorch:

from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained('batterydata/batteryscibert-cased')
model = BertModel.from_pretrained('batterydata/batteryscibert-cased')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)

And in TensorFlow:

from transformers import BertTokenizer, TFBertModel
tokenizer = BertTokenizer.from_pretrained('batterydata/batteryscibert-cased')
model = TFBertModel.from_pretrained('batterydata/batteryscibert-cased')
text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)

📚 Documentation

Model description

BatterySciBERT is a transformers model pretrained on a large corpus of battery research papers in a self-supervised fashion, starting with the SciBERT-cased weights. This means it was pretrained on the raw texts only, with no humans labelling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts.

More precisely, it was pretrained with the Masked language modeling (MLM) objective. Taking a sentence, the model randomly masks 15% of the words in the input then run the entire masked sentence through the model and has to predict the masked words. This is different from traditional recurrent neural networks (RNNs) that usually see the words one after the other, or from autoregressive models like GPT which internally mask the future tokens. It allows the model to learn a bidirectional representation of the sentence.

This way, the model learns an inner representation of the English language that can then be used to extract features useful for downstream tasks: if you have a dataset of labeled sentences for instance, you can train a standard classifier using the features produced by the BERT model as inputs.

Training data

The BatterySciBERT model was pretrained on the full text of battery papers only, after initialized from the SciBERT-cased weights. The paper corpus contains a total of 400,366 battery research papers that are published from 2000 to June 2021, from the publishers Royal Society of Chemistry (RSC), Elsevier, and Springer. The list of DOIs can be found at Github.

Training procedure

Preprocessing

The texts are tokenized using WordPiece and a vocabulary size of 31,116. The inputs of the model are then of the form:

[CLS] Sentence A [SEP] Sentence B [SEP]

The details of the masking procedure for each sentence are the following:

• 15% of the tokens are masked.
• In 80% of the cases, the masked tokens are replaced by [MASK].
• In 10% of the cases, the masked tokens are replaced by a random token (different) from the one they replace.
• In the 10% remaining cases, the masked tokens are left as is.

Pretraining

The model was trained on 8 NVIDIA DGX A100 GPUs for 1,000,000 steps with a batch size of 256. The sequence length was limited to 512 tokens. The optimizer used is Adam with a learning rate of 2e-5, $\beta_{1} = 0.9$ and $\beta_{2} = 0.999$, a weight decay of 0.01, learning rate warmup for 10,000 steps and linear decay of the learning rate after.

Intended uses & limitations

You can use the raw model for masked language modeling, but it's mostly intended to be fine-tuned on a downstream task. See the model hub to look for fine-tuned versions on a task that interests you.

Note that this model is primarily aimed at being fine-tuned on tasks that use the whole sentence (potentially masked) to make decisions, such as sequence classification, token classification or question answering. For tasks such as text generation you should look at model like GPT2.

🔧 Technical Details

• The model uses the Masked language modeling (MLM) objective for pre-training.
• It starts with the SciBERT-cased weights.
• The training data consists of 400,366 battery research papers from 2000 - June 2021.
• The texts are tokenized using WordPiece with a vocabulary size of 31,116.
• The model was trained on 8 NVIDIA DGX A100 GPUs for 1,000,000 steps with a batch size of 256.

📄 License

This model is released under the Apache-2.0 license.

Evaluation results

Final loss: 1.0505.

Authors

• Shu Huang: sh2009 [at] cam.ac.uk
• Jacqueline Cole: jmc61 [at] cam.ac.uk

Citation

BatteryBERT: A Pre-trained Language Model for Battery Database Enhancement