bert-base-uncased-embedding-step-scheduler

This model is a fine-tuned version of bert-base-uncased on the squad dataset.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('LLukas22/bert-base-uncased-embedding-step-scheduler')
embeddings = model.encode(sentences)
print(embeddings)

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2E-05
• per device batch size: 26
• effective batch size: 26
• seed: 42
• optimizer: AdamW with betas (0.9,0.999) and eps 1E-08
• weight decay: 1E-02
• D-Adaptation: False
• Warmup: False
• number of epochs: 3
• mixed_precision_training: bf16

Training results

Epoch	Train Loss	Validation Loss
0	0.0647	0.0876
1	0.0328	0.0826
2	0.0298	0.082

Evaluation results

Epoch	top_1	top_3	top_5	top_10	top_25
0	0.586	0.778	0.843	0.911	0.968
1	0.596	0.792	0.853	0.917	0.969
2	0.595	0.794	0.854	0.917	0.97

Framework versions

• Transformers: 4.25.1
• PyTorch: 1.13.1
• PyTorch Lightning: 1.8.6
• Datasets: 2.7.1
• Tokenizers: 0.12.1
• Sentence Transformers: 2.2.2

Additional Information

This model was trained as part of my Master's Thesis 'Evaluation of transformer based language models for use in service information systems'. The source code is available on Github.