Ish

@TheAhmadOsman Gold!

- you are - a random CS grad with 0 clue how LLMs work - get tired of people gatekeeping with big words and tiny GPUs - decide to go full monk mode - 2 years later i can explain attention mechanisms at parties and ruin them - here’s the forbidden knowledge map - top to bottom, how LLMs *actually* work - start at the beginning - text → tokens - tokens → embeddings - you are now a floating point number in 4D space - vibe accordingly - positional embeddings: - absolute: “i am position 5” - rotary (RoPE): “i am a sine wave” - alibi: “i scale attention by distance like a hater” - attention is all you need - self-attention: “who am i allowed to pay attention to?” - multihead: “what if i do that 8 times in parallel?” - QKV: query, key, value - sounds like a crypto scam - actually the core of intelligence - transformers: - take your inputs - smash them through attention layers - normalize, activate, repeat - dump the logits - congratulations, you just inferred a token - sampling tricks for the final output: - temperature: how chaotic you want to be - top-k: only sample from the top K options - top-p: sample from the smallest group of tokens whose probabilities sum to p - beam search? never ask about beam search - kv cache = cheat code - saves past keys & values - lets you skip reprocessing old tokens - turns a 90B model from “help me I’m melting” to “real-time genius” - long context hacks: - sliding window: move the attention like a scanner - infini attention: attend sparsely, like a laser sniper - memory layers: store thoughts like a diary with read access - mixture of experts (MoE): - not all weights matter - route tokens to different sub-networks - only activate ~3B params out of 80B - “only the experts reply” energy - grouped query attention (GQA): - fewer keys/values than queries - improves inference speed - “i want to be fast without being dumb” - normalization & activations: - layernorm, RMSnorm - gelu, silu, relu - they all sound like failed Pokémon - but they make the network stable and smooth - training goals: - causal LM: guess the next word - masked LM: guess the missing word - span prediction, fill-in-the-middle, etc - LLMs trained on the art of guessing and got good at it - tuning flavors: - finetuning: new weights - instruction tuning: “please act helpful” - rlhf: reinforcement from vibes and clickbait prompts - dpo: direct preference optimization — basically “do what humans upvote” - scaling laws: - more data, more parameters, more compute - loss goes down predictably - intelligence is now a budget line item - bonus round: - quantization: - post-training quantization (PTQ) - quant-aware training (QAT) - models shrink, inference gets cheaper - gguf, awq, gptq — all just zip files with extra spice - training vs inference stacks: - deepspeed, megatron, fschat — for pain - vllm, tgi, tensorRT-LLM — for speed - everyone has a repo - nobody reads the docs - synthetic data: - generate your own training set - model teaches itself - feedback loop of knowledge and hallucination - welcome to the ouroboros era - final boss secret: - you can learn *all of this* in ~2 years - no PhD - no 10x compute - just relentless curiosity, good bookmarks, and late nights - the elite don’t want you to know this - but now that you do - choose to act - start now - build the models

RAG systems return different context for the same query. This makes debugging impossible and outputs unpredictable. AvocadoDB guarantees determinism - same query, same context, verifiable by SHA-256 hash. - 100% deterministic (not probabilistic top-k) - Pure Rust, 40-60ms latency, local embeddings - 95%+ token utilization (vs 60-70% typical) - Exact file + line citations - $0 cost (no API required) Install: cargo install avocado-cli avocadodb.ai

The next billion-dollar founder has 15 followers on X rn. I will find you & fund you!

@OpenAI Been a pro user for months. Still no access. It just takes me to a subscribe Page.

We’re launching a research preview of Codex: a cloud-based software engineering agent that can work on many tasks in parallel. Rolling out to Pro, Enterprise, and Team users in ChatGPT starting today. chatgpt.com/codex

@hmalviya9 Would love to contribute. Please DM me the bank details!

Setting up an environment to study for these kids is more important than anything else in life right now.

@vladtenev @RobinhoodApp Linking family IRA/Roth accounts so irs easy to manage from one login.

What else do you want to see on @RobinhoodApp?

@saylor Why does this chart have El Salvador on the top right 👀?

I don't think this reflects what I got done last week.

🚀 Progress Towards AI-Driven Meta-Analysis What once took weeks can now be done in minutes. Exmplr AI's meta-analysis APIs empower agents to analyze and refine data with unprecedented speed. Accessible via our dev portal, they’re already transforming research. The future of research is here. 🌟 #DeSci #AI $EXMPLR #DeSciAI

Technical Product Managers are the perfect founder archetype for 2025. The bar for shipping a v1 has never been lower thanks to gen AI coding tools, and you have the exact skills needed right now: customer instinct + tech fluency. There's never been a better time to stop managing backlogs and start founding.

🚀 Dev Update: Training the Exmplr AI Agent for DeSci! $EXMPLR We are actively feeding API data to our AI agent to make interactions more engaging and insightful. Next, we'll integrate APIs for interventions & outcomes, pushing its ability to handle richer, structured queries for all personas. Future goal? Full-text study results for on-the-fly meta-analyses! #DeSci #AI #MedTech

🌍 Exmplr AI Expanded Access: Hope Meets DeSci 🚀 In our poll, 54.4% of you prioritized finding breakthrough treatments. With Exmplr Expanded Access, we’re empowering patients, families, agents, and researchers to discover life-saving solutions and drive innovation through shared knowledge. $EXMPLR 🌐 Find a life-saving treatment expandedaccess.exmplr.ai 💻 Fuel innovation developer.exmplr.ai/api/exmplr-exp… #Exmplr #DeSci

@primordialsense @sekoia_virtuals @wai_combinator @holyrektz @everythingempty @VirtualsWaifus @ExmplrAI @ai Open a support ticket with @virtuals_io discord server? We have no control over the LP.

@sekoia_virtuals @wai_combinator @holyrektz @everythingempty @VirtualsWaifus @ExmplrAI @ai ??? why not the team of @ExmplrAI @wai_combinator @eshbtc

Hey @everythingempty can we send you a @VirtualsWaifus wearing Virtuals gear? Sailor Waifu with @sekoia_virtuals tattoo. Could it get any better?

🚀 AI-Powered Precision EXmplr AI transforms clinical study exploration with AI-driven filters. Start with a broad search and narrow it down step by step to get exactly what you need—in seconds. $EXMPLR 🌐 Try it now: app.exmplr.io #Exmplr #AI #DeSci

From SaaS to Agents: The evolution is here. Replace subscriptions with micro-transactions, empower users with tokens, and align incentives with staking & governance. Decentralization isn’t just a buzzword—it’s the future of real-world asset agents. 🚀 #Web3 #DeFi #RWA @ExmplrAI $exmplr

Please check out our interactive demo. @eshbtc @kiranjyoth app.storylane.io/share/dkwzz2nj…

Hi Exmplr Community! 👋 Since launching our agent on Virtuals, the response has been amazing! 🚀 Our app, 2+ years in development, has high traffic affecting search, but we’re scaling fast and focusing on our HIPAA-compliant AI Agent Thank you for your patience! 🙏 #AI #DeSci

@aixCB_Vc Please take a look at @ExmplrAI

Happy New ATH 🎉 We are 💎🙌

@LgDoucet @mxmob1x @cryptobeliefguy @ExmplrAI @kiranjyoth @kiranjyoth is the Dev! @LgDoucet is not associated with this project! Hope that clears the air. Thank you!

@eshbtc @mxmob1x @cryptobeliefguy @ExmplrAI @kiranjyoth hey while you're here can you tell everyone that i am in fact *not* the dev on this one x.com/LgDoucet/statu…

Is it your Virtual token? Can you confirm this @LgDoucet app.virtuals.io/prototypes/0x6…

@LgDoucet @mxmob1x @cryptobeliefguy @ExmplrAI Not a test! It’s a utility token for an agent that integrates with our proprietary ML engine, designed to answer all questions about clinical trials and generate ad revenue for drugs. Announcement and litepaper coming soon! @kiranjyoth

@mxmob1x @cryptobeliefguy @ExmplrAI @eshbtc have they said anything though? or just another "test"

@alphaeazzy Confirming launch! Information updated on the website. Announcement coming soon! @kiranjyoth @eshbtc

@arjunkhemani A Guy in India is more accurate.

AGI (A Guy from India) will take over your job.