BERT Pros and Cons

The best part about Bert is its ability to generate relevant word embeddings and also understand the word context before the embeddings generation. Review collected by and hosted on G2.com.

The downside that I have felt about using Bert is the limit on the tokens that are fed to the generator; also its inability to generate embedding for inputs of different lengths. Review collected by and hosted on G2.com.

We utilize BERT for personalized sales strategies, aiming to address customer inquiries and concerns about products and services, thereby enhancing our customer service quality. This enables us to send tailored emails to customers with content relevant to their preferences and effectively target them with sales offers for products that align with their potential interests Review collected by and hosted on G2.com.

ChatGPT

I initially didn't perceive any drawbacks, but I did encounter some challenges at the outset. One significant issue was the inherent opacity of BERT, making it challenging to fully grasp the rationale behind its predictions.

Both the deployment and training phases incurred substantial computational costs, rendering it challenging to gain insights into the specific reasoning behind BERT's predictions.

Additionally, biases may exist in the model's output, owing to the utilization of extensive datasets during BERT's training. Review collected by and hosted on G2.com.

Bert is one of the best Large language model. I have used so far it solves many challenging use cases in my work Review collected by and hosted on G2.com.

accuracy of questions answering should be improved Review collected by and hosted on G2.com.

I think the most important part of this tool is that is reliable. I worked with ChatGPT for a while and I loved it, but it was crashing a lot. BERT so far has been a good tool when my other ones don't work. Review collected by and hosted on G2.com.

I think that this tool can feel a little more natural, what I mean with this is the responses, still feel a little "robotic" but if this can be changed I think the tools has potential. Review collected by and hosted on G2.com.

The versatile use with high accuracy in performance Review collected by and hosted on G2.com.

The model sometimes overfits on custom training and the steps for regularisation is bit complex Review collected by and hosted on G2.com.

Summarising, understanding text, sentiment or meaning of a text.easy to use. Review collected by and hosted on G2.com.

Accuracy can be increased, should try to match it's competition. Review collected by and hosted on G2.com.

BERT(Bidirectional Encoder Representations from Transformer) is basically a transformer model trained to work on NLP tasks such as sentiment analysis, text classification, Text summarization. It is trained on large dataset of arond 2500M words and hence its output is more accurate than other pre trained Hugging Face models. It can also handle polysemy means words with multiple meanings and homonymy means words that have same sounds but have different meanings which are two difficult problems with NLP tasks and I find it best configure and detect that words during my project duration. Review collected by and hosted on G2.com.

It needs used of high computational resources and it is trained on words only, so commonsense knowledge is lacking in it.Also it is biased on the training data which means sometimes efficient models may effect in production. Review collected by and hosted on G2.com.

BERT's foundation in Transformers is groundbreaking architecture. It makes dynamic interactions easier. Review collected by and hosted on G2.com.

I dont have anything that I do not like about bert. Review collected by and hosted on G2.com.

Bert gives the quick response compare to other AI tool ,it's memory requirements is low in Bert and accuracy is high and good too. And the best part it's free of cost. Review collected by and hosted on G2.com.

I faced one issue in Bert is that it's ability to handle long response is weak, sometimes server is slow, Other than that it's quite good tool it's improving day by day. Review collected by and hosted on G2.com.

Easy to use for any one and very efficient Review collected by and hosted on G2.com.

Time taken to train the model and sometime generalization. Review collected by and hosted on G2.com.