
Head of AI @RootcodeAI. Currently making machines more intelligent 🧠
Sri Lanka
Joined April 2020
- Tweets335
- Following770
- Followers123
- Likes1.8K
Thiru Dinesh retweeted
We officially achieved ISO/IEC 27001:2022 certification!
This recognition reflects our commitment to maintaining top-notch data security across all our operations.
Looking forward to continually strengthening how we protect data every day.
Thiru Dinesh retweeted
It was a pleasure to be part of the @rootcode_ai event on the future of AI hosted by Dubai Future Foundation.
Our Maitha Alsuwaidi discussed Microsoft's impactful AI initiatives about governance & growth and how we're pioneering AI strategies that shape the future.
Thiru Dinesh retweeted
Thirunayan, Head of AI at @rootcode_io will be speaking on how Enterprises can leverage potential of Generative AI at the AI Community Meetup - Dubai. He will share case studies on working with 112-year old manufacturing pioneer in Europe to leverage GenAI in their AI strategy.
Thiru Dinesh retweeted
The new 70 Q-bit quantum computer by @GoogleQuantumAI has performed a calculation that would take #Frontier (the largest supercomputer) 47 years. This is a massive milestone in scaling up Q-bits and bringing quantum computers for practical use.
#AI
thequantuminsider.com/2023/0…
Thiru Dinesh retweeted
One of the most underrated use-cases of AI is in optimizing supply chains. Not a fancy #LLM usecase but can definitely optimize global trade at a macro-scale reducing costs along with carbon footprint.
#AI #supplychain
Quick thought: The owner of the platform with the highest user engagement growth within the next quarter should get some form of advantage in the cage fight.
@elonmusk
#ElonVSzuck
Thiru Dinesh retweeted
London-based @glowb_uk announced it has raised €1.4 million led by Ada Ventures and joined by Active Partners, as well as Voyagers Climate-Tech Fund and Volta Circle, both early-stage sustainability investors 🇬🇧 🌞eu-startups.com/2023/06/lond…
Thiru Dinesh retweeted
Yesterday @OpenAI dropped one of the largest updates to it's APIs including a #GPT-3.5 variation with 16k context length, reduced prices and a new functionality called "Function Calling" which can potentially replace @langchain 's tools functionality.
openai.com/blog/function-cal…
Thiru Dinesh retweeted
We’re hosting our first AI Community meet-up in Colombo on the 20th June. You can meet me and the @rootcode_ai team there. Happening at @hatchonline from 6:30 pm onwards.
A great effort from the entire team and hoping to do much more in the future!! Extremely happy to say that we @rootcodeai have merged one of our largest contributions to @Rasa_HQ.
#AI
We just completed our largest #opensource contribution to date to @Rasa_HQ -one of the most used conversational #AI frameworks in the world.
Big shoutout to our AI team, who has been working for over 6 months on this! Here are a few snaps on how we celebrated this major milestone
The integration of ads into #chatbots is a major puzzle to solve for #bigtech. A year from now, we are either going to see a totally different revenue model for search engines or we are going to have chatbots sneak in ads in the middle of conversations
nytimes.com/2023/04/16/techn…
Thiru Dinesh retweeted
Have you heard about @karpathy's nanoGPT repo? It's making waves in the machine-learning community and is definitely worth checking out. It's just as impressive as @jnd1er's "The Design of Everyday Things" in the design community
#AI #LLM #GPT4
The train.py file in Karpathy's nanoGPT is a marvel. It does everything right, exactly the opposite of what you were taught in school. Global variables at the top, tons of top-level code (no main), a big while True
github.com/karpathy/nanoGPT/…
That would be the Uno reverse card of the decade
Can’t wait for ChatGPT to rug pull everyone and start charging $49/mo for access.
Thiru Dinesh retweeted
Can’t wait for ChatGPT to rug pull everyone and start charging $49/mo for access.
Thiru Dinesh retweeted
While most effort in the #AI #research is being put into improving the transformer architecture, parallelism is a core enabler to train large transformer models at scale. This paper by @MSFTResearch introduces a new library to do just that.
#pytorch
paperswithcode.com/paper/tor…
Thiru Dinesh retweeted
Very excited this has been accepted @LogConference and I can freely talk about what I have been during over the summer at @DeepMind. This is my last paper as a PhD student, and I am happy that the final chord was a major one. I had fun and learnt a lot from these amazing people!
A core aspect of intelligence seems to be learning trade-offs in a multi-agent setting. Could a #CICERO like model with multi-modal neurons and a physics view of the world be the path towards #AGI?
Would be interesting to know more viewpoints on this?@ID_AA_Carmack @ylecun
Replying to @ylecun @polynoamial
Since strategic communication and language manipulation is a major part of human intelligence, could a generalized multi-modal version of #CICERO be the key step towards #AGI ?
@sama @GaryMarcus
Thiru Dinesh retweeted
Strategic communication and negotiation is a task that even humans struggle to apply effectively, #CICERObyMetaAI is a fascinating model which has achieved human level performance in a game that is linguistically strategic.
Meta AI presents CICERO — the first AI to achieve human-level performance in Diplomacy, a strategy game which requires building trust, negotiating and cooperating with multiple players.
Learn more about #CICERObyMetaAI: bit.ly/3GBwLzx

















