Pranav

Be free, be true, be you

[𝘿𝙚𝙨𝙞𝙧𝙚 𝘾𝙤𝙙𝙚 @desirecode_]

The dark reality of modern marriage:

[Pranav @pranovaa_]

@rashem48 Nice. I'll try too

[Pranav @pranovaa_]

@rashem48 Solely through Opus 5.5 , or do you add any connector or something to the model for making video?

Can someone tell?

[Pranav @pranovaa_]

The people building the AI race are now asking us to slow it down.
It’s like a sweet maker telling us to eat less sugar. 😅

[sergio @sergiorocaa]

🚨 ¡ÚLTIMAS NOTICIAS! El ingeniero que creó Claude Code desde cero acaba de publicar un video de 28 minutos que es oro puro:

cómo escribir prompts que realmente entregan resultados absurdamente buenos.

Ya vi cursos de US$ 300 que ni siquiera se acercan a lo que él explica en los primeros 10 minutos.

Archivos CLAUDE.md, atajos de memoria, sesiones paralelas y patrones de prompting que casi nadie usa…

Todo en un solo video. Totalmente gratuito. Sin rodeos.

No importa si eres desarrollador, estás empezando ahora o ya usas Claude desde hace meses: esto cambia el juego para ti a partir de hoy.

Guarda. Mira. Comparte.

[kuno @kunoo]

You need to get angry.

[Pranav @pranovaa_]

- Stripe has agreed to acquire OpenRouter.
- Reports say NVIDIA to acquire Hugging Face for around $12.9 bn.

Let's see what happens when companies already playing major roles in payments & computing take positions in these platforms

#AI #NVIDIA #Stripe #OpenRouter #HuggingFace

[Aashu GJ @aashucoffee]

Hindustan Times is no more a newspaper.

[Pranav @pranovaa_]

True

[Pranav @pranovaa_]

Razorpay built Vulcan, an AI foundation model specifically for payments.

- Trained on ~3 trillion data points from 4B payments.
- It routes payments, detects fraud, assesses risk & predicts payment methods.
- Early results looks positive.

#AI #Fintech #Razorpay #Payments

[Pranav @pranovaa_]

Have you remember a movie's start and ending better than the middle? AI models do too.

It's called "lost in the middle" — LLMs pay less attention to info buried in the middle of long context window.

Thus, placement matters as much as content.
#AI #LLM #PromptEngineering #GenAI

[Pranav @pranovaa_]

LLMs don’t see letters. A tokenizer splits words into chunks (tokens) like “straw”+“berry”. The model works on tokens, not characters, so counting letters inside a token isn’t native.

#LLM #LargeLanguageModels #GenerativeAI #ArtificialIntelligence #AILimitations

[Pranav @pranovaa_]

We built RAG so the model would stop inventing facts.
It still made things up — ignored docs, invented details, even cited a source while contradicting it.

Retrieval alone doesn't fix hallucination.

#RAG #RetrievalAugmentedGeneration #LLM #GenerativeAI #AIHallucinations

[Pranav @pranovaa_]

We built RAG so the model would stop inventing facts.
It still made things up — ignored docs, invented details, even cited a source while contradicting it.

Retrieval alone doesn't fix hallucination.

#RAG #RetrievalAugmentedGeneration #LLM #GenerativeAI #AIHallucinations

[Pranav @pranovaa_]

LLMs don’t see letters. A tokenizer splits words into chunks (tokens) like “straw”+“berry”. The model works on tokens, not characters, so counting letters inside a token isn’t native.

#LLM #LargeLanguageModels #GenerativeAI #ArtificialIntelligence #AILimitations

[Nigel @nigeldesouza_]

This came to mind. Still hilarious to see.

[Pranav @pranovaa_]

@Bente254 @Wise1Philosophy is 6 not blocked?😅

[Pranav @pranovaa_]

@wiseconnector 'Unlimited Money' bcuz, then we won't have to do something for money which means we get unlimited free time. And that time can be used to gain knowledge. hope i've interpreted the ques well. 🙂