GlowLy

Software Engineer at @vllm_project

[LIFE AI @LifeNetwork_AI]

1/ Life AI Testnet is coming!
We’re opening access to the first OG members who will help shape the future of personalized, proactive healthcare.
OG Role = Exclusive benefits, early access, and special rewards throughout our journey.
Here’s EXACTLY how to secure your spot 👇

[Andrej Karpathy @karpathy]

Excited to release new repo: nanochat!
(it's among the most unhinged I've written).

Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single, dependency-minimal codebase. You boot up a cloud GPU box, run a single script and in as little as 4 hours later you can talk to your own LLM in a ChatGPT-like web UI.

It weighs ~8,000 lines of imo quite clean code to:

- Train the tokenizer using a new Rust implementation
- Pretrain a Transformer LLM on FineWeb, evaluate CORE score across a number of metrics
- Midtrain on user-assistant conversations from SmolTalk, multiple choice questions, tool use.
- SFT, evaluate the chat model on world knowledge multiple choice (ARC-E/C, MMLU), math (GSM8K), code (HumanEval)
- RL the model optionally on GSM8K with "GRPO"
- Efficient inference the model in an Engine with KV cache, simple prefill/decode, tool use (Python interpreter in a lightweight sandbox), talk to it over CLI or ChatGPT-like WebUI.
- Write a single markdown report card, summarizing and gamifying the whole thing.

Even for as low as ~$100 in cost (~4 hours on an 8XH100 node), you can train a little ChatGPT clone that you can kind of talk to, and which can write stories/poems, answer simple questions. About ~12 hours surpasses GPT-2 CORE metric. As you further scale up towards ~$1000 (~41.6 hours of training), it quickly becomes a lot more coherent and can solve simple math/code problems and take multiple choice tests. E.g. a depth 30 model trained for 24 hours (this is about equal to FLOPs of GPT-3 Small 125M and 1/1000th of GPT-3) gets into 40s on MMLU and 70s on ARC-Easy, 20s on GSM8K, etc.

My goal is to get the full "strong baseline" stack into one cohesive, minimal, readable, hackable, maximally forkable repo. nanochat will be the capstone project of LLM101n (which is still being developed). I think it also has potential to grow into a research harness, or a benchmark, similar to nanoGPT before it. It is by no means finished, tuned or optimized (actually I think there's likely quite a bit of low-hanging fruit), but I think it's at a place where the overall skeleton is ok enough that it can go up on GitHub where all the parts of it can be improved.

Link to repo and a detailed walkthrough of the nanochat speedrun is in the reply.

[GlowLy @glowly29]

RT @durov: Telegram and xAI have agreed to a 1-year partnership to distribute Grok to Telegram’s billion+ users and integrate it into its a…

[Elon Musk @elonmusk]

Tesla vs other electric car companies

[Jeff Dean @JeffDean]

Someone just reminded me of this lecture I gave in 2009 that described the evolution of Google Search from 1999 to 2009. People who are interested in how our search systems work might find this interesting.

It touches on disk-based serving systems, in-memory indices, compression schemes for inverted indices, latency issues due to interference from background processes, queries of death, evolution of hardware, and more..

Video: https://t.co/OnFk1azDk9
Slides: https://t.co/q4WZXRWQg6

[GlowLy @glowly29]

@JeffDean @vdean314 cool

[Orbit Labs @orbit__labs]

🚀 Hello #LUNC Community!

Following the successful approval of the v3.4.0 software upgrade proposal (#12157), we are now actively working on the Wasmd unforking as our next priority.

Link below: 👇

📌 What’s Next?
After completing the Wasmd unforking, our next major milestone will be upgrading to Cosmos SDK 50. This upgrade will bring significant improvements to the Terra Classic ecosystem.

Stay tuned for more updates! 🚀🔥

[Orbit Labs @orbit__labs]

We at OrbitLabs are reviewing the Tax2Gas Pull Requests to assist with moving things forward. We're committed to support LUNC chain development. 🚀🚀🚀
#LUNC #Tax2Gas

[Orbit Labs @orbit__labs]

🚀 Exciting news for #TerraClassic! We're thrilled to introduce OrbitLabs, a dedicated blockchain development team ready to elevate $LUNC to new heights! 🌖 @trinhkien99 @tienorbitlabs @HoaBka @TropicalDog2

[GlowLy @glowly29]

I just claimed my spot for @MachidotXYZ liquidity bootstrapping event. Come check your eligibility at https://t.co/lM50ay9kt0

[GlowLy @glowly29]

Agree

[pixel @spacepixel]

Last chance before banana zone

[GlowLy @glowly29]

@0xBreadguy #FRIEND @friendtech ?

[Đỗ Hùng Việt @dohungviet]

The entire nation mourning the loss of its leader - Comrade Nguyen Phu Trong, General-Secretary of the Communist Party of Viet Nam 🇻🇳

[Avail @AvailProject]

Mainnet is coming. Turn on your notification.

23-07-24

[GlowLy @glowly29]

@Sickde_One @obumnwabude I have same error. It works when I deploy, test on localnet. But I got error when run on testnet/devnet

[GlowLy @glowly29]

Getting some Testnet SOL at @quicknode's faucet for my vtUp3Mbg wallet!
https://t.co/1qp9kKIKZ7

[toly 🇺🇸 @toly]

https://t.co/ao8vJjetGm

[Andrew Huang ⚡️⛓️ @KAndrewHuang]

Chains in service of great apps — love to see some of the best builders in the space get it!

[GlowLy @glowly29]

@0xLuxeCrypto buy #friend bro

[GlowLy @glowly29]

@dj_d_sol @hdevalence me too? how many commits are enough?