Hacker News: nandakishor_ml

New comment by nandakishor_ml in "I built non-autoregressive decision models with RL a year ago"

nandakishor_ml — Sat, 19 Sep 2026 15:21:28 +0000

just not these jev guys not opensourcing their work

New comment by nandakishor_ml in "I built non-autoregressive decision models with RL a year ago"

nandakishor_ml — Sat, 19 Sep 2026 14:58:09 +0000

I didn't claimed it bro it was first. Just shared the findings here.

New comment by nandakishor_ml in "Laya the open source version of Jev"

nandakishor_ml — Sat, 19 Sep 2026 14:44:45 +0000

I built a pypi for it called hallunox https://pypi.org/project/hallunox/

New comment by nandakishor_ml in "I built non-autoregressive decision models with RL a year ago"

nandakishor_ml — Sat, 19 Sep 2026 14:43:44 +0000

It's property benchmarked btw. And do read the og paper at https://arxiv.org/abs/2510.01237 And btw pypi package also there which never mentioned https://pypi.org/project/hallunox/

New comment by nandakishor_ml in "Laya the open source version of Jev"

nandakishor_ml — Sat, 19 Sep 2026 14:41:35 +0000

All the stuff are proper benchmarked. Feel free to read the paper, https://arxiv.org/abs/2510.01237

New comment by nandakishor_ml in "I built non-autoregressive decision models with RL a year ago"

nandakishor_ml — Sat, 19 Sep 2026 14:39:08 +0000

The paper is one year old. https://arxiv.org/abs/2510.01237 https://pypi.org/project/hallunox/

New comment by nandakishor_ml in "I built non-autoregressive decision models with RL a year ago"

nandakishor_ml — Sat, 19 Sep 2026 10:46:58 +0000

This project was built on the exact research on jev architecture research one year ago

I built non-autoregressive decision models with RL a year ago

nandakishor_ml — Sat, 19 Sep 2026 10:46:58 +0000

Article URL: https://laya.convaiinnovations.com/

Comments URL: https://news.ycombinator.com/item?id=49765348

Points: 1215

# Comments: 292

New comment by nandakishor_ml in "Open-sourced jev architecture last year with model,paper and dataset"

nandakishor_ml — Thu, 17 Sep 2026 13:41:01 +0000

We are human , frustration is something we feel when same kinda architecture is closed sourced and we celebrate it

New comment by nandakishor_ml in "Open-sourced jev architecture last year with model,paper and dataset"

nandakishor_ml — Thu, 17 Sep 2026 05:13:49 +0000

Feeling that open-source community do not get the recognition they deserve

Open-sourced jev architecture last year with model,paper and dataset

nandakishor_ml — Thu, 17 Sep 2026 05:13:00 +0000

Everyone now talks about the architecture that's not auto regressive and does lightning fast probability prediction with a json schema. I worked on this literally one year back in March 2025, published an arxiv paper, pushed the model to huggingface along with the pypi package and training dataset. And then one year later, a frontier lab came, proposing the same idea like literal breakthrough without technical papers, open weights and no open dataset. For anyones information the main guiding model is RL not embedding model or LLM

Paper: https://arxiv.org/abs/2503.23303

Model: https://huggingface.co/DeepMostInnovations/sales-conversion-model-reinf-learning

Dataset: https://huggingface.co/datasets/DeepMostInnovations/saas-sales-conversations

Also the second work published in September 2025 was exactly the same one jev proposed now

Paper: https://arxiv.org/abs/2510.01237

My model uses PPO over sequence embeddings to output turn-by-turn conversion trajectories (probabilities from 0.0 to 1.0).

Jev uses parallel sampling (trained via RLCD) to output confidence distributions and schema choices.

It's incredibly frustrating that the thing that you made with months of hard work, sweat and sleepless night is architecturally similar with the vertical use case and don't get the support you deserve because frontier lab build something horizontal. The open-source story in general

Comments URL: https://news.ycombinator.com/item?id=49736660

Points: 82

# Comments: 13

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Tue, 15 Sep 2026 08:11:24 +0000

You do need to sign DPA with them

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sun, 13 Sep 2026 03:46:57 +0000

Healthcare btw and also the llm that done the code is bad at review and if you do write code people who don't want to send credentials to cloud

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 19:20:40 +0000

not without sending your code to them, if you have .env file with variables exposed, you will need to rotate again.

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 19:04:04 +0000

the whole vision is not sending code to cloud servers to audit. So people dont worry about the egress

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 19:02:06 +0000

thats nice suggestion tbh

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 18:39:32 +0000

its a need of the hour

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 18:07:58 +0000

But you do need a product for sales aint it

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 18:06:49 +0000

So going for an incubator is a good approach as I do have functioning product

New comment by nandakishor_ml in "How SaaS startup guys get first 100 customers first, make fkn $500k ARR fast?"

nandakishor_ml — Sat, 12 Sep 2026 18:04:31 +0000

but sending 300 ish emails make all of my email outbound to spam right!!