Re: AI/Machine Learning, Bharat and Bhartiya IT Industry
Posted: 20 Jan 2023 01:36
We should feed Siachen thread archives into ChatGPT.
Consortium of Indian Defence Websites
https://forums.bharat-rakshak.com/
2 hours long.We build a Generatively Pretrained Transformer (GPT), following the paper "Attention is All You Need" and OpenAI's GPT-2 / GPT-3. We talk about connections to ChatGPT, which has taken the world by storm. We watch GitHub Copilot, itself a GPT, help us write a GPT (meta !) . I recommend people watch the earlier makemore videos to get comfortable with the autoregressive language modeling framework and basics of tensors and PyTorch nn, which we take for granted in this video.
it completes the sequence with the outcome and so it's a language model in that sense nowI would like to focus on the under
the hood of um under the hood components of what makes chat GPT work so what is the
neural network under the hood that models the sequence of these words and that comes from this paper called
attention is all you need in 2017 a landmark paper
why just siachen, why not entire BRF???Vayutuvan wrote:We should feed Siachen thread archives into ChatGPT.
I'll go one step further.ArjunPandit wrote:A short sighted and potentially unintended consequence of chatgpt could be that data will be more held privately. More countries will be asking asking to localize the data and companies will have to pay to the nation states. China India have taken such actions in past. There will be rise of local companies. China already has a head start in this. Days of free access to social media for all are few. Others could follow the suit.
Another of course will be reduction in knowledge economy jobs: This includes
1. IT and college educated jobs.
2. Basic uni jobs will be going away.
3. Youtube will be the uni and perhaps you will have to pay for it, given the traffic/volume or it would shift to tiered viewership models where first few minutes are free and then paid
on a lighter note pinchar jobs will still require humans for some time
AN economist, about 20 years ago, had suggested that the profits made on manufacturing based on AI should be used to provide minimum pay to the unemployed.titash wrote:..........
6. The other barrier to AI is legislation. No one can trump the vote of the unemployed. Driverless cars in India anyone?
ArjunPandit wrote:why just siachen, why not entire BRF???Vayutuvan wrote:We should feed Siachen thread archives into ChatGPT.
Why write any judgements in English in the first place? They can write in an Indian language and machine-translate into Angrezi.Cyrano wrote:SC has been experimenting with machine translation of its judgements in English into several languages. CJI Chandrachud spoke about it at an India Today event recently. He roped in an Indian start up IIRC. Good move, can be scaled up and implemented across the country.
Milards are pukka angrezVayutuvan wrote:Why write any judgements in English in the first place? They can write in an Indian language and machine-translate into Angrezi.Cyrano wrote:SC has been experimenting with machine translation of its judgements in English into several languages. CJI Chandrachud spoke about it at an India Today event recently. He roped in an Indian start up IIRC. Good move, can be scaled up and implemented across the country.
This has been the norm, where there are "distilled"models accompanying larger models. Definitely something unique to Alpaca 7B.sanman wrote:
Any way, the bigger picture here is that the really Large Language Models can be used to train smaller ones (like a mother training her young?)
This could lead to the creation of various more specialized language models, meant to serve smaller niches.
I think last week's announcement of Alpaca 7B from Stanford is the reason why Elon Musk and his fellow Tech Gurus are speaking out about AI right now. Seeing that AI is now spreading out to the masses via these knowledge-transfer/distillation processes has made them fearful that a genie is now out of the bottle. Meh, I don't see how they can hope to curtail or regulate it. Govts will continue researching it, underground hackers and hobbyists will continue doing it. You can't constrain coding - that's absurd.ernest wrote:This has been the norm, where there are "distilled"models accompanying larger models. Definitely something unique to Alpaca 7B.
When BERT came out kinda ushering in the age of LLMs, it was accompanied by DistilBERT that was smaller and could run in PCs. "Knowledge DIstillation"and "Student Teacher Curriculum Learning" are active areas and the norm in some ways.
IBM to Pause Hiring for Jobs That AI Could Do
Roughly 7,800 IBM jobs could be replaced by AI, automation
CEO Krishna says IBM to pause hiring for replaceable roles