This model is based on anferico/bert-for-patents - a BERT
LARGE
model (See next section for details below). By default, the pre-trained model's output embeddings with size 768 (base-models) or with size 1024 (large-models). However, when you store Millions of embeddings, this can require quite a lot of memory/storage. So have reduced the embedding dimension to 64 i.e 1/16th of 1024 using Principle Component Analysis (PCA) and it still gives a comparable performance. Yes! PCA gives better performance than NMF. Note: This process neither improves the runtime, nor the memory requirement for running the model. It only reduces the needed space to store embeddings, for example, for semantic search using vector databases.
BERT for Patents
BERT for Patents is a model trained by Google on 100M+ patents (not just US patents).
bert-for-patents-64d huggingface.co is an AI model on huggingface.co that provides bert-for-patents-64d's model effect (), which can be used instantly with this prithivida bert-for-patents-64d model. huggingface.co supports a free trial of the bert-for-patents-64d model, and also provides paid use of the bert-for-patents-64d. Support call bert-for-patents-64d model through api, including Node.js, Python, http.
bert-for-patents-64d huggingface.co is an online trial and call api platform, which integrates bert-for-patents-64d's modeling effects, including api services, and provides a free online trial of bert-for-patents-64d, you can try bert-for-patents-64d online for free by clicking the link below.
prithivida bert-for-patents-64d online free url in huggingface.co:
bert-for-patents-64d is an open source model from GitHub that offers a free installation service, and any user can find bert-for-patents-64d on GitHub to install. At the same time, huggingface.co provides the effect of bert-for-patents-64d install, users can directly use bert-for-patents-64d installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
bert-for-patents-64d install url in huggingface.co: