HNHacker News
TopNewBestAskShowJobs

okpatil

21 karma · joined August 12, 2013

submissionscomments
okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
That's an absolutely fair way. More power to you !
okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Don't assume. It has generic world knowledge. It is performing well on the benchmarks we didn't even train it on.
okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Of course yes. They will be inside the executable.
okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
The same engine is is available as an evaluation API.

curl -s -X POST https://at0m.pienomial.com/decide/v0 \ -H 'Content-Type: application/json' \ -d '{ "state": "Charged twice for the same card payment this morning.", "questions": { "queue": {"type":"choice", "instructions":"Which team should handle this?", "criteria": {"billing":"invoices, charges, refunds", "technical":"outages, bugs, deploys", "fraud":"unauthorised or suspicious activity"}}, "urgent": {"type":"noul", "instructions":"Needs action today."}}, "email_id": "you@example.com"}'

If it fits your use case, you are welcome to use it.

When it is a rust standalone rust executable, as it is powering the API, it becomes just plug and play. No dependencies needed.

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Actually it was created around 12 years ago. We just launched the model around 3 days ago.

Apologies if it is too much of a bother.

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Point taken. Let me explain if you allow me.

It is possible with deterministic decision models, such as At0m, to gauge the probabilities at every decision. This behavior in addition to hard coded logic, it is possible to completely replicate a prompt's logic.

Using Fable 5.1, it is a matter of minutes.

I believe that most of the compliance check documents will be a solved problem, 3-6 months in future.

None of the LLMs can do it.

Hence I asked to the comment poster if he would want to demo, so that I can show it to him, how to do it step by step. By bad, if it came out too strongly.

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Good point.

To counter, most of the AI is not open. So is none of Microsoft Products. As long as they work, we keep using them.

okpatil··on Clef: Open-source decision models, and new RL fine-tuning platform
We were able to completely automate 20,100 token prompts with At0m[https://at0m.pienomial.com/].

We believe entire compliance workflows (even multilingual) could be automated.

Would you like to get a demo ?

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
We got you. How would you use it though ? Through Apps ? Would love to discuss.

At0M: A 60M local Jev at 16 ms latency and 79% accuracy on Typed Decision

https://at0m.pienomial.com/ https://news.ycombinator.com/item?id=49920350

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
We got you.

At0M: A 60M local Jev at 16 ms latency and 79% accuracy on Typed Decision

https://at0m.pienomial.com/ https://news.ycombinator.com/item?id=49920350

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Atom is 60M Param (around 133x to 400x smaller).

16ms latency. And locally run.

https://at0m.pienomial.com/

Why go big when you can go small ?

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
It is a single rust executable, with model embedded inside of it. Current evaluation API is being run by the same.

We wanted to stress test the system before the V1 release.

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
32K
okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
It's open for benchmarking. V1 release next week.

Businesses are built on outliers. It doesn't make sense throwing your hard earned insights while paying them money to steal it.

Also, cloudflare https://robindev.substack.com/p/cloudflare-took-down-our-web...

okpatil··on Clef: Open-weight decision models, and new RL fine-tuning platform
Why give cloudflare your data ?

At0M: A 60M local Jev at 16 ms latency and 79% accuracy on Typed Decision

https://at0m.pienomial.com/ https://news.ycombinator.com/item?id=49920350

okpatil··on India’s second wave of Covid-19 feels nothing like its first
Hello, fellow Kolhapuri here. Maybe we can connect.
okpatil··on Show HN: Terrastruct – A Diagramming Tool for Systems
We have developed something on these lines. Please send me an email omkar@wynum.com for a demo.
okpatil··on Superdeterminism: The path we didn’t take
I am just going to put it here. This wikipedia article is about an ancient Indian school of thought which believed that everything is composed of atoms, but the aggregation and nature of these atoms was predetermined by cosmic forces. They also believed in absense of "Free Will". https://en.wikipedia.org/wiki/%C4%80j%C4%ABvika
okpatil··on Facebook Messenger Platform
Spot on !!
okpatil··on Why I Still Use Python for High Performance Scientific Computing
Just typed "import antigravity" in python repl. It took me to the same link. :-)
okpatil··on Material Design Lite Components in HTML/CSS/JS
It might be brand "GOOGLE" that is getting them excited
okpatil··on Ask HN: Who wants to be hired? (June 2015)
Location : Brisbane

Remote : Yes

Willing to relocate : No

Technologies : Golang, Nodejs, Angularjs, Rails, Meteorjs, Cordova, Ionic Framework, Mongodb, Redis

Resume/CV : https://www.dropbox.com/s/dnmqnznsjov6mmr/Omkar%20Resume%20%...

Email : okpatil4u@gmail.com

okpatil··on How India's Patent Office Destroyed Gilead's Global Game Plan
> I don't see where they show that the cost of clinical trials etc. for getting drugs through the FDA is affected. From what I can tell, those are the largest costs involved. Especially considering that most of the drugs discovered, even using those advanced methods, fail in the trial phase after hundreds of millions invested.

Agreed, but I compared the costs of developing drugs 20 years ago. They had to synthesize and experiment with each and every drug. Not that's not the case. And most of the drugs which being developed now, use the principals I explained above. If clinical tests are proven successful then they file patents, but it's not an "innovative" process as per say. India has been working to get rid of such patent trolls for some time now.

> "Siphon money" or "capture the rewards of the value they provide"? Undermining global incentives to invest billions in research that could save hundreds of millions around the world does not seem like the "right direction". Not the first time the Indian government has done something selfish but shortsighted. (Pokhran comes to mind.)

Agreed, these kinds of researches has to be done, but not at cost of people's lives, I have explained statistics of people affected by AIDS above. It is a grave social injustice. These research and development processes are flawed in the sense that they cannot provide affordable alternatives to existing methods. These companies must adapt to affordable and agile development processes or someone else will replace them, remember "Mars Orbiter Mission" by ISRO.

And speaking of Pokhran, it was not a short sighted decision. India has Pakistan, an irresponsible nuclear capable neighbor, with whom India has been in state of war since 1947. Effects of Pokhran economic sanctions were not severe, because foreign trade of India constituted to 4% of it's GDP, while USA participation was only 10%. After Pokhran, under then PM Atal Bihari Vajpayee, India achieved record breaking 6-7 % growth rate in GDP. A few problems could be pointed out, but it was not a short sighted decision at all.

I am not being a socialist, but I must add that such innovation and development should be done for the people, not American people, not British people, but keeping people of the world in mind. Please don't monopolize products which can save a lot of peoples lives.

okpatil··on How India's Patent Office Destroyed Gilead's Global Game Plan
You don't understand gravity of situation. India has third highest population of AIDS patients. Around 2.1 million people are infected[1]. That is twice the population of Cyprus. Around 170k children and adults lost their lives in 2009 [2]. India has 32.7 % of its population below poverty line [3]. That's an income below $1.25 per day. If they can't pay for the drugs doesn't make them less deserving.

And as far as drug developments go, due to GPU computing and molecular dynamics and monte carlo simulation methods, the cost of developing drugs has gone significantly down [4][5]. For starters they take a molecular structure, add or replace a functional group, and run the simulation with a particular protein structure. Drug development costs are not that high now, compared to twenty years ago. Most of the money goes in marketing.

If Government of India thinks more about it's 2.1 million than bunch of corporate companies trying to siphon money from wherever they can, then YES, they are in the right direction.

1. http://www.bmj.com/content/340/bmj.c621 2. http://www.indexmundi.com/g/g.aspx?c=in&v=37 3. http://povertydata.worldbank.org/poverty/country/IND 4. http://www.ncbi.nlm.nih.gov/pubmed/16758486 5. http://www.biomedcentral.com/1741-7007/9/71

okpatil··on India Orders 32 Websites Blocked, Including GitHub, Archive.Org, Pastebin
Being an Indian, I wish that this was true.
okpatil··on Show HN: Initial release of H2O, and why HTTPD performance will matter in 2015
It seems that everything mentioned in the library could be done with golang easily. I am interested to see how H2O benchmarks with pure golang binaries.
okpatil··on Why India outsourcing is doomed
This statement is plain wrong. People you are referring to are highly educated professionals.
okpatil··on Ask HN: Appropriate toy for a 1-year old girl that may lead to programming
Poor girl..
okpatil··on Another go at Go failed
>Ruby or nodejs for web servers You haven't tried go, have you?
okpatil··on Compiling Rust for GPUs
I don't think you read the thread properly. It's mentioned here, https://groups.google.com/d/msg/golang-nuts/8OJ6etdl6WY/0FKM..., that you could access cuda architecture with golang. And as mentioned here http://commandcenter.blogspot.it/2012/06/less-is-exponential..., golang was designed to replace c++. Any modern language which gives flexibility for both high and low level use would live, if you don't think so, go and advertise assembly. If rust gives cuda and http-server, one could easily design a project like this http://www.rescale.com.
Page 1 of 2Next →