Research Graph

Connecting Scholarly Communications

[Amir Aryani @amir_aryani]

RAG: Supercharging AI with Smart Data Retrieval https://t.co/cXe1slWiUx

[CADREAUS @cadreaus]

Platform to assess applications to access sensitive data. Safe person? Correct data for correct project?
https://t.co/15TjonLLKv

[Amir Aryani @amir_aryani]

Great to see Reproducible Outputs as one of the new functions of GPT-4 https://t.co/SvRLdCe7Pe #gpt #artificalintelligence

[Amir Aryani @amir_aryani]

15 New Initiatives in the New White House Order on AI
https://t.co/y4DnZqRWOk

[SYC @SYCa000]

While they are effective in understanding and generating natural language, they have limitations such as producing incorrect responses, known as the hallucination effect. This can lead to potential risks in clinical settings, especially in areas like imaging appropriateness

[SYC @SYCa000]

Large language models (LLMs) like ChatGPT, developed by OpenAI, have shown success in various tasks due to their advanced architecture and training mechanisms.
#aigraph
https://t.co/oS4txQew9e

[Amir Aryani @amir_aryani]

Mistral AI is 187x cheaper compared to GPT-4
https://t.co/GcpgEsEiCn

[YunLearnAI @YunLearnAI]

Feature engineering refers to the process of creating new input variables from the available data. #aigraph
https://t.co/UC3m6J08Pi

[YunLearnAI @YunLearnAI]

data preparation can be treated as another hyperparameter to tune as part of the modeling pipeline.

https://t.co/Fbhsghd5wD

[YunLearnAI @YunLearnAI]

Transformers, introduced in 2017, are a pivotal architecture for LLMs, designed around the concept of attention, which helps process longer text sequences efficiently

https://t.co/lQLz92QFDy
#LLM #NLP

[SYC @SYCa000]

Google Bard will destroy ChatGPT because it’s a long-term game, and Google has the upper hand in terms of market dominance. ChatGPT has a better product, but that doesn’t mean they’ll win the long race.
#aigraph https://t.co/0JP5XW2HvF

[SYC @SYCa000]

This post details how to apply three core UX concepts to generative AI products: 1) affordances, 2) feedback, and 3) constraints.
#aigraph
https://t.co/k2y1BI6sr6

[Aishwarya Nambissan @AishLearnsAI]

Engineers can now test models without worrying about privacy concerns since generative AI can produce large amounts of synthetic data that mirrors real-world information
#aigraph

https://t.co/ZPLTBfrkro

[SYC @SYCa000]

Large language models have found applications in a wide range of fields, transforming the way we interact with technology and enabling new possibilities, including NLP, Chatbots and virtual assistants, automation, Language translation, etc.
#aigraph
https://t.co/eBnUCpUhdn

[Aishwarya Nambissan @AishLearnsAI]

Dimensionality reduction results in fewer input variables which allows for a simpler predictive model that may have better performance when making predictions on new data.
#aigraph

https://t.co/4oTbUknXjk

[SYC @SYCa000]

The output of the OCR process is a text that contains typos, recognition mistakes, non-text symbols, and other inaccuracies.
Another challenge that machine learning engineers face is what to define as a word like Chinese, Japanese, or Arabic.
#aigraph
https://t.co/DhNjrHZU8H

[Aishwarya Nambissan @AishLearnsAI]

LLMs cannot understand written text directly, so Sentence Embedding is carried out. Thanks to the large dimension of the vector created by embedding, small variations in the data can be seen with greater precision.

#aigraph
https://t.co/q6Bw1pko49

[Shuyi @Shuyi817]

With just a simple, but powerful calculation like the Cosine Similarity, we can build a reasonable recommendation system.
#aigraph
https://t.co/ROEAamxno2

[YunLearnAI @YunLearnAI]

Applying LLM on the keyword research will be a revolution of seo. #aigraph

https://t.co/5LHqIWXlbM

[SYC @SYCa000]

LLMs have many specific types.
Autoregressive language models, Transformer-based models, Encoder-decoder models, Pre-trained and fine-tuned models, Multilingual models, and Hybrid models. For instance, LLaMa2 we are using is transformer-based.
#aigraph
https://t.co/O9mqJ2OiCB