Meet Shah @Curiousmeet

@ai4bharat @iitmadras Are there plans to add models from India like sarvam chat sutra etc. in here because that would be awesome.

Last night I taught nanochat d32 how to count 'r' in strawberry (or similar variations). I thought this would be a good/fun example of how to add capabilities to nanochat and I wrote up a full guide here: github.com/karpathy/nanoc… This is done via a new synthetic task `SpellingBee` that generates examples of a user asking for this kind of a problem, and an ideal solution from an assistant. We then midtrain/SFT finetune on these to endow the LLM with the capability, or further train with RL to make it more robust. There are many details to get right especially at smaller model sizes and the guide steps through them. As a brief overview: - You have to ensure diversity in user prompts/queries - For small models like nanochat especially, you have to be really careful with the tokenization details to make the task easy for an LLM. In particular, you have to be careful with whitespace, and then you have to spread the reasoning computation across many tokens of partial solution: first we standardize the word into quotes, then we spell it out (to break up tokens), then we iterate and keep an explicit counter, etc. - I am encouraging the model to solve the model in two separate ways: a manual way (mental arithmetic in its head) and also via tool use of the Python interpreter that nanochat has access to. This is a bit "smoke and mirrors" because every solution atm is "clean", with no mistakes. One could either adjust the task to simulate mistakes and demonstrate recoveries by example, or run RL. Most likely, a combination of both works best, where the former acts as the prior for the RL and gives it things to work with. If nanochat was a much bigger model, you'd expect or hope for this capability to more easily "pop out" at some point. But because nanochat d32 "brain" is the size of a ~honeybee, if we want it to count r's in strawberry, we have to do it by over-representing it in the data, to encourage the model to learn it earlier. But it works! :)

Today we launched Tinker. Tinker brings frontier tools to researchers, offering clean abstractions for writing experiments and training pipelines while handling distributed training complexity. It enables novel research, custom models, and solid baselines. Excited to see what people build.

Thinking Machines @thinkymachines

9 months ago

Introducing Tinker: a flexible API for fine-tuning language models. Write training loops in Python on your laptop; we'll run them on distributed GPUs. Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!

240 790 6K 4.2M 3K

After an incredible response in Labs, we’re starting to roll out AI Mode in Search to everyone in India (English to start). It’s a total reimagining of Search, and we’re excited for even more people to use it.

happy July 4th! Ollama is celebrating by building.

@AskPerplexity how many dra centers does Microsoft google and Amazon have in India

@grok @AskPerplexity @AskPerplexity still no reply from you

@AskPerplexity can you generate a video of a person generating electricity from heat coming out of the data center.

📣 Announcing Delta Lake 4.0.0! We are excited to announce the release of Delta Lake 4.0! 🎉 This major release is packed with powerful new features and improvements designed to make your lakehouse experience even better. 🚀 Release Highlights 🌟 𝗣𝗿𝗲𝘃𝗶𝗲𝘄: 𝗖𝗮𝘁𝗮𝗹𝗼𝗴-𝗠𝗮𝗻𝗮𝗴𝗲𝗱 𝗧𝗮𝗯𝗹𝗲𝘀 Native support for catalog-integrated lakehouse tables—laying the foundation for unified governance and discoverability. 🌟 𝗗𝗲𝗹𝘁𝗮 𝗖𝗼𝗻𝗻𝗲𝗰𝘁 𝗳𝗼𝗿 𝗦𝗽𝗮𝗿𝗸 𝗖𝗼𝗻𝗻𝗲𝗰𝘁 Delta Connect is an extension for Spark Connect which enables the usage of Delta over Spark Connect, allowing Delta to be used with the decoupled client-server architecture of Spark Connect. 🌟 𝗩𝗮𝗿𝗶𝗮𝗻𝘁 𝗗𝗮𝘁𝗮 𝗧𝘆𝗽𝗲 Support for the Variant data type to enable semi-structured storage and data processing, for flexibility and performance. 🌟 𝗜𝗻𝘀𝘁𝗮𝗻𝘁 𝗗𝗥𝗢𝗣 𝗙𝗘𝗔𝗧𝗨𝗥𝗘 Remove table features without truncating your table’s history or requiring downtime. Delta Lake 4.0 is a major leap forward for the open lakehouse community, offering advanced management, greater flexibility, and modern architecture. 🔗 Check out the official release notes: github.com/delta-io/delta… #opensource #oss #deltalake #linuxfoundation

New short course: DSPy: Build and Optimize Agentic Apps DSPy is a powerful open-source framework for automatically tuning prompts for GenAI applications. In this course, you'll learn to use DSPy, together with MLflow. This is built in partnership with @databricks and taught by @ChenMoneyQ, co-lead of the DSPy framework. Many AI builders spend hours hand-tuning prompts. When given a set of evals, DSPy automates this process. It’s especially useful for optimizing prompts, including few-shot prompts, in complex agentic AI workflows. Further, if you switch an application to a newer LLM, performance can degrade if your prompts were optimized to the previous model. DSPy automatically optimizes the entire system for the new LLM as well, using just a few evaluation examples. This course teaches DSPy works, and best practices for using it. You’ll write programs using DSPy’s signature-based programming model, debug them with MLflow tracing -- to gain visibility into how different parts of a pipeline, as well as how the overall system, are performing -- and automatically improve their accuracy with DSPy Optimizer. Please sign up here: deeplearning.ai/short-courses/…

In the wake of the recent terrorist attack in Pahalgam, the urgency to fortify India’s borders—both physical and digital—has never been clearer. It was an absolute honor to participate in the AIFSS 2025, sharing the stage with the Hon’ble Attorney General of India, Shri R. Venkataramani, and contributing to the visionary roadmap laid out by the Hon’ble Home Minister, @AmitShah Shri Amit Shah. 🙏 TWO AI team and I personally remain deeply committed to harnessing the power of AI to strengthen India’s cybersecurity infrastructure and national resilience. Ministry Of Home Affairs (mha), GOI National Forensic Sciences University (NFSU) TWO AI Grateful to the Ministry Of Home Affairs (mha), GOI and National Forensic Sciences University (NFSU) for the opportunity and warm invitation.

NUMERIC (TWO AI) @numericmachines

a year ago

TWO AI CEO Pranav Mistry (@pranavmistry) addressed delegates at #AIFSS2025 in New Delhi, highlighting the role of AI in strengthening India's cybersecurity. #AI #AIFSS #SUTRA by TWO AI

1 9 78 32K 4

@grok can you answer this

@AskPerplexity How can llm accomodate large contexts from delta tables?

@grok can you answer this

@AskPerplexity , when will we have Identity insert in delta tables in Azure Synapse/fabric spark?

@AskPerplexity , tell me good resources to learn pdf data extraction using llm with lower hallucinations?

Both #SUTRA models SUTRA-V1 and SUTRA-R0 leads inference cost performance benchmarks (Tokens/word) beating #GPT #Deepseek #Llama and almost all others.

NUMERIC (TWO AI) @numericmachines

2 years ago

SUTRA is #1 in Indian Language AI 🇮🇳 🌟 A recent independent study evaluating tokenization across India's 22 official languages has shown that SUTRA's tokenizer outperforms other Large Language Models (LLMs), including those specifically designed for Indic languages. SUTRA

1 9 24 7K 7

Today, we are infusing the power of agentic AI into the GitHub Copilot experience, elevating Copilot from pair to peer programmer 🤖 (1/4) github.blog/news-insights/…

find it here: arxiv.org/abs/2501.09223