Akshay Bharde
1.6K posts
@avbharde
STOP THE WAR NOW!!!
Lost my primary X account @AI_Feb21. Secondary account to stay connected in the meantime. Pls amplify this msg to help me reach my network till i get back of my account. Thank you @unseenvalue @_chartitude @InvestmentBook1 @suru27 @LearningEleven @ishmohit1 @Prakashplutus
I have to push back a little here specifically on Mythos, Fable, and the framing around Anthropic's models. All of these, including Anthropic's Claude family, are built on transformer architecture. There's no fundamental architectural breakthroughs here that India is locked out of. The transformer is public knowledge. Sarvam, Krutrim, and others are building on it too. So the question isn't just whether we can access these models it's whether we should keep chasing the same architectural paradigm at all. And my honest answer is: we shouldn't. We should stand on our own and go further. At Blankline, we've been working on something we call Hope a post-transformer research initiative. The core thesis is that transformers compute the wrong probability operation for general intelligence. They estimate P(next token | context). What intelligence actually requires is P(latent structure | observations), followed by search and verifier-driven reasoning. That's a fundamentally different thing. Our Hope-1 architecture discrete-latent program codes, verifier-driven search, a pre-registered seven-rung validation framework has already cleared four of those seven rungs at 0.69M to 3M parameters, exceeding the closest published baseline by approximately 2× on ARC tasks. The research isn't about building a better LLM. It's about asking whether the whole paradigm is wrong. This matters for what you're saying. If you're right that technology is now the ultimate sovereignty question, then copying the transformer race whether from American or Chinese open-source models still leaves India downstream. What changes the equation is a research bet that could shift the cost frontier of general intelligence entirely. That's what we're working toward. I turned down funding that would have made Blankline instantly credible overnight. Those investors wanted US and UK IP transfer as a condition. If I were just building a product, I might have taken it. But the research we've done and what we have internally right now is something I wasn't willing to hand over. The value isn't in the model weights. It's in the architectural thesis and the research direction itself, which we've published openly at blankline.org/research/hope while keeping the implementation proprietary. India's top investment firms I've spoke with wants profits, not pure R&D. That's a real problem and I've run into it directly. The SF ecosystem understands how transformative foundational research can become. We haven't built that culture here yet but that's exactly the culture India needs if we want to be a source of frontier intelligence, not just a consumer of it. The goal isn't to make intelligence available only to Indians. It's to make it available to everyone and that only happens if someone builds the thing from first principles rather than licensing access to someone else's architecture.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
My protein diet is as under- I get up at 5am- 1- take a black with soaked nuts like walnuts, almonds, and pumpkin seeds- 5-7 grams protein 2- Go to Gym at 6-6.15, come back at 7.30- take 3-4 boiled eggs- 18-24 grams protein. 3- 9.15-9.30 breakfast- Dal chilla with Paneer/Greek Yogurt with chia seeds, hemp seeds and one fruit/soya or paneer tikki total around 25 grams protein. Lunch at 12.30-45- Curd, salad, paneer, one vegetable/pluses- around 25-30 gram protein. Dinner at 6.30- Chicken/fish/paneer/soya/Greek yogurt/ anything protein rich with some sorted vegetables/very less quantity of rice. 25-30 grams.
Visited a nano-cap SME in the aluminium windows and façade business yesterday. Came back with a lot of respect for what happens behind the numbers. As analysts, we spend most of our time looking at revenue growth, margins, ROCE and cash flows. But when you visit a company, you realize that numbers are just the end result of years of hard work. A few things I learnt: 1. Many management efforts don't show up in one year's numbers. Some decisions take years before the results become visible. 2. There are a lot of challenges being handled every day that investors never see. We often make judgments sitting in front of a screen without fully understanding what is happening on the ground. 3. Many small family-run businesses may not have the best systems, controls or investor presentations. But that doesn't automatically mean there is a lack of integrity. However, intent to improve should be there. In fact, I was surprised to see how involved and passionate the entire family was about the business. 4. One interesting thing was that the promoters were genuinely worried about investor communication because of the fear of SEBI regulations. Many SMEs want to communicate better with shareholders but need the right guidance on how to do it. Plant visits don't always change your investment thesis. But they definitely help you understand the people behind the business. And sometimes that context is as important as the numbers themselves. Disclosure: Alphawealth and I don't have any investment in the company.
@Shardhr There are no answers to these. Many may not like this but Eventually you get your 100 baggers by surrendering to the market and circumstances.
