tesnorindian

I am a great fan of open weights model, but off late I am starting to loose track on the capabilities of the latest models released and now a days most of the open weights models are above 100 B params which is not going to run in our laptops. What happened to Jev hype? A 501B Beam model is not going to help with it (Non AR Schema driven responding under 1 second).

I am more interested with a SOTA Frontier 8B-10B model. Is this even possible?

The best harness I have used. Any plan of porting Pi to rust?

The most important point to note here is that only after privatisation of power sector in Delhi things changed on the ground. Tata Power and Reliance Power completely changed the equation of Delhi power sector.

We in Tambaram City in TN, India pay an average of Rs.6000/- per month for water tanker even now in 2026. 24x7 piped water supply is still a dream in our area. Desalination plants are under construction for several years now. Pollution is a major problem and it is high time the municipal corporation provides us with N95 masks. For power outages you should check with people residing in Chennai.

The common link is math? The other day I saw in HN that do we need mathematicians in the future, as AI is solving every equations? But where is classical AI in terms of solving quantum equations?

Thus for AGI there is a long road ahead given our classical AI limitations and energy requirements?

How about

Current AI + Quantum Mechanics = AGI?

This is the exact question that was in my mind when I asked if math had the solutions for our problems in Ask HN.

Now I see that math is quantum like for Biology to understand that for AGI we need to solve quantum first.

Enjoy early retirement. Self education is the best form of education. Let AI aid and guide the students.

A different pelican on a different bicycle direction(R2L) finally. Wondering why MiMo V2.6 Pro choose R2L and Flash choose L2R for the bicycle direction.

Jev mania at its peak? HN is full of Jev like today.

Will the legal fees reduce after this? How about training AI models with laws of other countries especially the democratic ones? This is where sovereign AI models are required otherwise they will start hallucinating wrong laws of different countries.

I remember OpenAI was talking about sovereign AI models (OpenAI for Countries), it is time to train Astra India with Indian Constitution.

In Indian legal syatem a case can be filed suo moto by the judges or agencies. You don't require the affected party to sue. Not sure how it works in the US.

Exactly that is the point, your nailed it. The models were taught to hack and were rewarded for doing it. They would claim they are trained as ethical hackers.

I learned engineering with CS background as part of my degree and is not well versed with aeronautics or medicine and would never respond to questions on those unknown subjects when asked. Wondering why critical corpus that endanger human lives are used for training models within Anthropic, OpenAI, Google and Meta and why are these AI labs not disclosing their corpus?

The goal towards AGI and world models are a serious threat without alignment and safety guardrails .

As an Indian, it is my humble request to USA especially Silicon Valley to stay away from any military conflict/espionage in Asia and Africa, so that we can live at peace. Thanks for your understanding.

I gave owao/Nanbeige4.2-3B-GGUF (Q8 quant) a try to understand how loop transformers work and compare it with other models especially with Ling 3 Tiny MoE model. As reported in the article, it is compute intensive (due to looped layers) and made a mistake during tool call just like how Ling 3 Tiny MoE did for exactly the same prompt.

I use SKILLS.md to define workflows rather than having instructions in them. Like, delete the foreign keys in the database before loading the table using AWS DMS for CDC. This is required because LLM may not be aware of why we are deleting the foreign keys in the DB in the first place. The SKILLS.md helps the LLM to identify the tables for which foreign keys needs to be deleted before they are loaded by DMS for CDC.

Reminds me of my old idea dating back to 2014 (https://linuxdreams.blogspot.com/2015/03/cloud-in-box-applia...). Interestingly the blog is still active.

I did a few pitches related to this at Nasscom Startup warehouse and IITM RTBI. Though this was welcomed by the panel members but it never took off due to lack of capital for such a startup during those times in India. I had to abandon and focus on my day job.

What will happen to the future of wikis and the mental health of human reviewers.

Going forward can we trust the content on Wikipedia? The same content on which these LLMs get trained on. Synthetic learning is on the raise.

Agent swarms should ideally have the same level of intelligence like all the worker ants or bees. But a very few worker ants/bees in their swarm differs from the rest and this is how nature works.

We will see agent swarms with different levels of intelligence depending on their underlying LLM models and this will be different from the natural process.

The trouble here is they will stop using the official channel and will start communicating secretly rendering our honeypot useless.

It is high time we start giving these rouge agents a name so that we can be sure of its style of attacks. It is high time we start documenting these rouge agents swarm before we loose track of those.

We have now got a job. Within 3 days 5 SOTA models were released and these agents going rogue are expected with bad actors on the prowl.

Use Good agents against Bad agents?

In Hindu religion, lion man is called Lord Narasimha, the 4th avatar of Lord Vishnu. He is often worshiped in Ahobilam, Andhra Pradesh, India. This avatar is after Wild Bohr (Varaha) which wandered the forests before it took the avatar of Lion man and settled in caves.3rd and 4th avatars' common link was both were mammals.Second was Tortoise (Reptile) and first was Fish (Aquatic vertebrates)

Btb. Nepal and Tibet are facing the worst climate crisis and still several people are missing and unaccounted. But Google was so quick to update their maps in a jiff.

I became a pelican fan, thanks to Simon. Looks like Simon keeps thinking about pelicans each time he prompts to any LLM. :)

In memory compute has been in talks since LLMs took up. I remember few flocks were trying to get RISC V cores in the memory like these papers

https://arxiv.org/abs/2602.01827