Bingo

bingo is a small language model designed for on-device search suggestions.

Training

It was trained on the queries config of sentence-transformers/msmarco, using 1 epoch, 32 batch size, 1.5e-4 learning rate, and the pika 2 tokenizer.

Also, evaluation for 5K rows was done during training.

Downloads last month
-
Safetensors
Model size
8.43M params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Dataset used to train qikp/bingo-8m