Skip to main content

This Blog and the Tools Used (+ how to build one of your own)

 



I've tested many dozens of AI tools and features during the research and writing of this blog. I'd like to have utilised even more, if I had a powerful enough graphics card to download and execute different LLMs then I would have. My Dell workstation though is limited by 32GB and a 4GB NVIDIA Quadro card with ageing Xeon processor. My Dell laptop has 16GB, a 12th Gen i7 and 4GB 3050. These are averagely powerful pieces of hardware, but are insufficient to run much in the way of Open Source AI models, at least ones that I'd use as regular tools. 

I state this to demonstrate that most people's usage of AI tools is likely to be through 'cloud' services, such as the new (clunky) ChatGPT4 Apple OS phone app. Which is a shame, as it does hinder people's understanding of how these tools work, and what sort of libraries they are dependent upon. This may ease up, as more smaller and capable LLMs become available, but it's probably a bit too late. Why is this of any importance at all? Well I'd suggest that a semblance of understanding how these tools work is of benefit to comprehending their strengths and weaknesses.

For those interested there follows a list of the tools I've used in preparing for, and writing this blog. Not all are AI.

  • Research and writing time: circa 83 hours per month. Without using the tools below, this would add a significant amount of time to the process. Have the AI tools themselves increased my efficiency by the claimed 40%, as often touted? Certainly, especially in searching for new white papers etc. Chat PDF has been very useful too, once I have downloaded a PDF.
  • Research tools used: Elicit, Consensus, Google Scholar, DuckDuck Go, 
  • Preparation Tools: YouTube Transcript, Chat PDF, Perplexity, 
  • Visualisation Tools: DALL·E, via Bing, DreamLike, 
  • Podcast Tool: Hindenburg Pro
  • Writing Tools: Notepad ++, LibreOffice, Initially also Bard for writing the initial background posts to this blog. Bard is also used, on occasion, late at night to correct grammatical mistakes when I'm just too tired to be bothered!
There remain few useful AI based text summarisers available, all are very limited in the amount of text that can be input (I expect this to change as the token size of LLMs increase) and, to be frank, many do an adequate job at best. 

Code your own tools


For those interested in installing and using their own PDF chat, and have also gain an understanding of how such tools work in practice, than this new video from Prompt Engineering will prove invaluable.




It would be good if and when more tools become available for researchers, of a higher quality. Automated source referencing within Word Processors, such as found in Perplexity would be a very useful addition, as an example. 

Tools I want to try out

Pi, Notion AI, Superus, DocuAsk, which may solve the size of file limitations I am currently experiencing? Adobe Firefly, SoundDraw, and finally Descript. which I've been looking at for a while now, but still haven't go around to using it. 

What sort of tools would you propose? What tasks could narrow AI help best with in your own blog / paper / article writing, research?

Labels:

Comments

Post a Comment

Popular posts from this blog

The AI Dilemma and "Gollem-Class" AIs

From the Center for Humane Technology Tristan Harris and Aza Raskin discuss how existing A.I. capabilities already pose catastrophic risks to a functional society, how A.I. companies are caught in a race to deploy as quickly as possible without adequate safety measures, and what it would mean to upgrade our institutions to a post-A.I. world. This presentation is from a private gathering in San Francisco on March 9th with leading technologists and decision-makers with the ability to influence the future of large-language model A.I.s. This presentation was given before the launch of GPT-4. One of the more astute critics of the tech industry, Tristan Harris, who has recently given stark evidence to Congress. It is worth watching both of these videos, as the Congress address gives a context of PR industry and it's regular abuses. "If we understand the mechanisms and motives of the group mind, it is now possible to control and regiment the masses according to our will without their
Post a Comment
Read more

Beware the Orca, the challenge to ChatGPT and Palm2 is here

  So Google's 'we have no moat' paper was correct. If you train an LLM wisely then it's cost effective and cheap to produce a small LLM that is able to compete or even beat established, costly LLMs, as Microsoft has just found. It's another excellent video from AI Explained, who goes through some of the training procedures, which I won't get into here. Orca, is a model that learns from large foundation models (LFMs) like GPT-4 and ChatGPT by imitating their reasoning process. Orca uses rich signals such as explanations and complex instructions to improve its performance on various tasks. Orca outperforms other instruction-tuned models and achieves similar results to ChatGPT on zero-shot reasoning benchmarks and professional and academic exams. The paper suggests that learning from explanations is a promising way to enhance model skills. Smaller models are often overestimated in their abilities compared to LFMs, and need more rigorous evaluation methods. Explana
Post a Comment
Read more

What is happening inside of the black box?

  Neel Nanda is involved in Mechanistic Interpretability research at DeepMind, formerly of AnthropicAI, what's fascinating about the research conducted by Nanda is he gets to peer into the Black Box to figure out how different types of AI models work. Anyone concerned with AI should understand how important this is. In this video Nanda discusses some of his findings, including 'induction heads', which turn out to have some vital properties.  Induction heads are a type of attention head that allows a language model to learn long-range dependencies in text. They do this by using a simple algorithm to complete token sequences like [A][B] ... [A] -> [B]. For example, if a model is given the sequence "The cat sat on the mat," it can use induction heads to predict that the word "mat" will be followed by the word "the". Induction heads were first discovered in 2022 by a team of researchers at OpenAI. They found that induction heads were present in
Post a Comment
Read more