paari_7

1/ Most people think AI document generation is solved. Just ask ChatGPT, right? Wrong. Try asking it to create a 20-page market analysis report with real data, proper citations, and executive formatting. You'll get generic fluff that took you 2 hours to prompt-engineering

@jackzhu1304 Sorry I didn't get you..can you please elaborate

@happened_7 I think your new data-related agent project might need a memory storage function.

1/ Most people think AI document generation is solved. Just ask ChatGPT, right? Wrong. Try asking it to create a 20-page market analysis report with real data, proper citations, and executive formatting. You'll get generic fluff that took you 2 hours to prompt-engineering

@noah_vandal Gemma and gpt-5.4-mini

@happened_7 what models?

Built a self-improving data agent over a 275-table MySQL DB using DSPy RLM + GEPA. No schema dumping, After 752 rollouts it answers complex multi-hop SQL questions cold. Demo dropping soon

@rachbelaid @LakshyAAAgrawal just raw, 275 tables of operational data, no derived models.

@happened_7 @LakshyAAAgrawal Did your DB had derived data models? Or just raw data models

@rachbelaid @LakshyAAAgrawal trying to drop it this weekend, would love if you could try it and share your results :)

@happened_7 @LakshyAAAgrawal Now I know where my weekend will be spent. Built a few analyst agent and convinced that it’s possible to build an agent that discovers and learn over time

@st_tronn Just Dm'ed you!:)

@happened_7 been working on something simillar, will love to connect dm me

@arattml @LakshyAAAgrawal Just DM’ed you :)

@happened_7 @LakshyAAAgrawal Dm me

@VedicWisdom1 Love marriage is wrong according to vedas??

People think Inter-Caste marriages cause Varnasankarta. Varnasankara are actually the people who 1. Commit adultery 2. Commit incest 3. Abandon duties of their Varna. So a Brahmin who does not do the occupation of Brahmin is Varnasankara. This is the original meaning of Varnasankara. Later in medieval times caste system came up, people designated certain castes and Varnasankara and inserted Shlokas in some Dharmashastras accordingly. Groups like Shakas, Hunas, Chinese all were included.

@original_ngv Opus 4.6

Ok real question. Which is better for writing? Sonnet 4.6 or Opus 4.6?

@aravind I was a student of the same degree

@WeekendInvestng @grok is it true, verify the fact

25% hike. Inflation says Hello !!

@himanshustwts @Alibaba_Qwen also launched their vector db

This is going to be Chinese week and we already have: > Deepseek V4. > GLM 5 > Minimax M2.5 What am i missing?

@xeophon @rasdani_ On learning to compact: One of the big advantages of having a *program* / forward function abstraction, not an agent abstraction, is that you'd just write two modules. The main multi-step ReAct and the compacter, and both would get RL’d/PO’d in your program’s forward function.

@original_ngv Bro is back

this would be great for single parents. usually one parent is the harsh one, and the other is the fun one. with this the single parent doesn't have to be both. the AI can do the hard stuff.

Hi @Hetzner_Online, your verification process is very slow and sometimes doesn't work at all. I wanted to get the web hosting service and tried both my driver's license and Aadhaar card, but your system failed to verify ... leading to frustration and me deleting my account.

Open sourced Qwen3-TTS voice generator. 105s audio <3min, 3 voices. Customize voices at app.py:46-96. "Joe Rogan vibes" gets you that voice. "Witty comedian" → comedy voice. Self-hosted Modal. No APIs. Apache 2.0. Thanks @Alibaba_Qwen github.com/paarijat007/qw…

@dx8152 @Ali_TongyiLab @ModelScope2022 It's really crazy stuff, tried it yesterday x.com/i/status/20153…

Qwen3-TTS has been officially released as open source, voice design, and voice cloning. I've created a custom WebUI interface for your convenience @Ali_TongyiLab @ModelScope2022 Tutorial: youtu.be/1dV2VTJXwo8 ⬇️github.com/hero8152/Qwen3…

@DataChaz Tried this,crazy stuff x.com/happened_7/sta…

Text-to-speech is moving way too fast. Just a few days ago, I tweeted about PersonaPlex-7B, NVIDIA's new open source TTS (x.com/DataChaz/statu…) And today, Qwen just open-sourced Qwen3-TTS 🤯 It’s a revolutionary text-to-speech model built for control. Not just about generating speech, but about shaping how it sounds directly from language. You can guide the pace, the tone, and the expressiveness straight from text, without touching audio graphs or hand-tuning parameters. That’s the real shift! What makes Qwen3-TTS stand out is how practical it already is: → voice cloning from just a few seconds of audio → voice creation without any reference sample → support for 10 languages out of the box → end-to-end latency down to ~97ms → works in both streaming and non-streaming setups The models come in two sizes (0.6B and 1.7B), so you can trade off quality and hardware cost depending on your setup. You can work with curated voices, designed voices, or cloned ones, and it integrates cleanly with vLLM for production use. It also ships as a simple Python package you can pip install. If you’re building real-time voice systems, this removes a lot of friction! 100% free and open source. I put the repo in the 🧵↓