Kris

Ngl This is heavily used in ingestion pipelines. If you have millions/thousands of docs to do ingestion. You'd have to use queues and workers to manage the pipeline 🙂

@devnp2007 Which year are you?

@kris_aieng Can you guide what should we focus on to crack interviews kind of projects we need ? How to get internships ?

Current offer I have in hand is in BLR 🥲 Is whitefield a good place to start staying in BLR?? - how are the expenses there? - how is the housing there? - what about transport?? how much the overall expenses come upto??

@ChinmayKak Got it

@kris_aieng I think either would be good, I read the PI blog first

Interesting blog on how RLMs as harnesses are better at subgoal decomposition, and thus lead to better generalisation! RLMs at the core handle long context by delegating tasks and by programmatic context offloading via Python REPLs and subagents, thus they can reduce complex problems -> smaller problems that are In distribution to transformers, thus helping compositional generalisation. Also worth checking the original RLM paper, Prime-intellect's blog on it as well:)

@sinhaniik Thanks Nikhil

1. Is Whitefield a good place to live? Yes. It’s a good place to start if your office is nearby. 2. How are the expenses? Compared to Indiranagar, Koramangala, or Hebbal, Whitefield is relatively more affordable. Your biggest expense will be rent. 3. How is the housing? Finding a good rental can take some time. Most landlords ask for 3–4 months’ security deposit. Before signing the agreement, ask how deductions from the deposit are handled when you move out. This is important—I had to pay repainting charges when I vacated my flat. I was paying around ₹15,000/month for a 1 BHK (no sharing). 4. What about transport? Connectivity is good. BMTC buses and the Metro make commuting fairly convenient. I mostly preferred public buses. The Metro is convenient, but if you use it daily, the cost adds up. 5. Overall monthly expenses? Around ₹25k–30k/month, depending on your lifestyle and rent.

@AmoghSaxena17 Got it. Still have sometime to look for other jobs

@kris_aieng At least! If u want to live in a flat, else u can go for double sharing PG

@AmoghSaxena17 Minimally it covers around 30k 🥲

@kris_aieng Descent rent as compare to rest of blr Metro connectivity is good Dead vibe If you prepare your meals or get a cook, expected 18k rent + 10k food + 10-30 recreational(subjective)

@Palakonweb nice portfolio & background has nice festive vibe

Uhm finally my portfolio is live. Link in the comments.

How come we don't see these ai videos in cricket?? Where is the avg cricket fan present right now???

@v0xium Starting with Triton is always better option when learning

CUDA C and Triton are both used to write GPU programs but they operate at very different levels of abstraction. CUDA C is a low-level extension of ansi C (now also C++) that gives us direct control over GPU threads, memory, and synchronization. So there is a lot of flexibility but then it comes with the cost of increased complexity and a lot of manual optimization. Triton is a Python based language (DSL) made for writing high performance tensor ops. Here we describe computations on tensor blocks while the compiler handles many low level optimizations on its own. That is why Triton is much easier to write. So I would say that CUDA is the better choice when maximum control and flexibility are required while Triton is good for rapidly developing kernels. Try to use both, see which one fits your use case.

Recursive Language Models (RLMs) Read here: outcomeschool.com/blog/recursive…

@deedydas Might reach billions as they have enough GPUs

Moonshot (Kimi) is now well over $300M and approaching 400

Chinese AI labs are doing serious numbers. I looked up all the reported and leaked numbers of the 5 big independent labs: Moonshot (Kimi), DeepSeek, Zhipu (GLM), Kuaishou (Kling), Minimax and the 2 hyperscaler labs: Bytedance (Seedance) and Alibaba (Qwen). The private labs are doing $2.6B in revenue run rate. Still behind the big 2 labs, but 4 of 5 are in the top 25 of global AI companies by revenue. With Kimi K3 dropping and Seedance 2.5 coming soon, China continues to close the gap on LLMs and extend their lead on video. The AI race just does not stop. Disclaimer: some of the numbers on growth, revenue, gross margin and valuations are either being sought, rumored and purported by third party sources and not straight from the company. For example, DeepSeek numbers come from the Information and Zhipu from a Macquarie analyst. The public ones are listed in the Hong Kong exchange where only half year financials are mandatory, not quarterly.

@Palakonweb @kitto749277 It definitely worked to some extent and started again 🥲

@kitto749277 Go bald

If someone know how to stop hairfall kindly share your advices 😭😭

Bro is too humble for the content he created for cs336 🫡

@Palakonweb My laptop fan got fucked and had to pay 4k for it 😭

Gareeb ka to aata bhi gila hogaya aur gila aata bhi chori hogaya. My laptop charger got fu*ked had to pay 1k for it out of 2k bank balance.

@rasbt Much needed one 🫡

How can an LLM switch between low-, medium-, and high-effort reasoning? And how does an LLM learn to reason more or less? I put together a “little” article explaining how these effort levels are implemented at inference time and during training.

Right now at the openai build week hackathon at hyd It's always good to connect with people across different roles & age groups Vibe is good, let's see how it ends 🫡 @KushalVijay_ #OpenAIBuildWeekHyd

@aiwithanu Don't do that please consult someone before trying it

@Krishna70284154 oh okk will check that out too thankuuu

If you have never heard about sinusitis, you truly are very lucky. I guess I have suffered enough to even write a research paper on it🥲 If anyone has home made remedies, do drop it in the comments. Currently I take : turmeric + black pepper + lemon water + ginger powder.