Thilak Rao @Thilak Twitter profile

Last Seen Profiles

@OtraCR

@Sleeepy22

@__schanky__

@KasualZenkai

@88rafusRa

@oekaki_kesu

@mollyllewis

@MingZhong29826

@maki1815255

@__o2you

@chesney_moore1

@imyeowoooo

@apodo1043

@Meta_Mari_

@bader222abha

@rinsan157971

@AllFactstho

@i_8sai6

@MixonTownBamm

@johan1176060

@NeedMoreCourage

@Ctd5jBY2H7AC24y

@SuppachokP

@Shuck_A_Cuck

@CambridgeDay

@vrpancho

@wooOpi_

@nicominardi

@bomaasoma

@Mamamaui77

@rudevero

@bescoinyes

@eloocz

@heatherisone

@maxcelcat

@AlkanResit02

Thilak Rao

@Thilak

5 months

I knew @levelsio was a grifter who’s real product was the dreams that he sells to all the indiehackers. TIL that he’s also an insecure loser

111

31

1K

Thilak Rao

@Thilak

4 months

> I got Llama 3 running in my browser using only my GPU with my Wi-Fi switched OFF completely client-side. Steal OSS software, never credit back and claim to have invented it yourself. Lie lie lie some more. Completely on-character for an MBA LARPing as an Indie Hacker. There’s

31

30

610

Thilak Rao

@Thilak

5 months

Just got Llama 3 8B Instruct by @Meta up and running on my iPhone with @private_llm , running fully on-device on my iPhone with a full 8K context on 8GB devices. Coming to all iPhones with 6GB or more RAM this weekend.

10

35

275

Thilak Rao

@Thilak

5 months

2

18

271

Thilak Rao

@Thilak

5 months

I’m all for AI. I built AI products myself (and these aren’t API wrappers). But next token predicticting LLMs (aka “AI”) are no substitute for a trained therapist. There are multiple cases of people who need human help but instead went with an AI chatbot and were driven to

16

11

237

Thilak Rao

@Thilak

4 months

@mattshumer_ Sounds like you’ve never read the Apache 2.0 License. It requires attribution. And don’t even get me started on the models he’s been cribbing without attribution, off of CivitAI for his photos thing.

web-llm/LICENSE at main · mlc-ai/web-llm

High-performance In-browser LLM Inference Engine . Contribute to mlc-ai/web-llm development by creating an account on GitHub.

github.com

3

138

Thilak Rao

@Thilak

5 months

ChatGPT is great and all, but have you experienced the wild ride that is @Paytm ’s chatbot?

5

8

115

Thilak Rao

@Thilak

5 months

A 200 page book is ~55,000 tokens. It’s unlikely that an average user even uses half of this. Even if a user ends up chatting 24/7, it’ll be in the order of half a million tokens a month, which costs $0.69 a month. Is the $9.3 that @levelsio makes off of each user worth the risk

4

2

103

Thilak Rao

@Thilak

5 months

@levelsio For all the hate that big LLM like @OpenAI , @AnthropicAI and Google get, they’re right to stay as far away from ethically slippery usecases like this as possible. If he were using any of their services for this, he’d have been perma banned from using their APIs in no time. This

2

89

Thilak Rao

@Thilak

5 months

Two bit huckster! Cannot refute by criticism of his grifts because it’s true, so resorts to slandering my employer. Classic case of pot calling the kettle black. Good riddance!

8

2

84

Thilak Rao

@Thilak

4 months

Link for the lazy:

0

1

73

Thilak Rao

@Thilak

6 months

Just dropped @private_llm v1.7.7 for macOS! 🎉 Added 3 cool LLMs - Google Gemma 2B IT, BioMistral-7B & Nous-Hermes-2-Mistral-7B-DPO. Privacy-first, as always. Check it out!

Private LLM

@private_llm

6 months

v1.7.7 of Private LLM for macOS has just been released. This release introduces 3 new downloadable LLMs, all quantized with 4 bit OmniQuant quantization.🧵

1

2

12

3

5

38

Thilak Rao

@Thilak

3 months

@ElijahSchaffer Please don't visit India. We’d rather not have Nazi sex offenders who can't tell the difference between India and Pakistan.

The Blaze Fired Right-Wing Podcaster After Sexual Assault Accusation

Schaffer’s firing came after a weeks-long investigation into an allegation he had groped a colleague’s breasts without her consent, a source with knowledge told The Daily Beast.

www.thedailybeast.com

1

4

31

Thilak Rao

@Thilak

4 months

🎉 Huge update for @private_llm on macOS! v1.8.5 brings the 4-bit OmniQuant quantized Llama 3 8B Instruct model (h/t @AIatMeta ) and all its fine-tunes, such as the amazing Llama 3 Smaug 8B 🐉(h/t @abacusai and @bindureddy ), to Mac users. If you're a Mac user, you won't want to

Private LLM

@private_llm

4 months

v1.8.3 of Private LLM for iOS was just released! 🎉 Check out the new models in this release: 🧵

1

2

25

1

5

34

Thilak Rao

@Thilak

4 months

@jameswlepage Dear @levelsio bootlicker, I hope you realise that he’s merely posting a cropped out screengrab of the webllm website. Wow! What a grand experiment, in plagiarism?! 🤣

1

0

31

Thilak Rao

@Thilak

6 months

@bindureddy Mandatory government approval for GenAI models? There goes innovation taking a backseat for the foreseeable future. What a way to put the brakes on AI in India. A sad day for progress, indeed.

2

4

33

Thilak Rao

@Thilak

3 months

@bindureddy Apple might suck at AI, but they also offer the best hardware and bang for the buck for running large LLMs offline. NVIDIA tho…

9

2

32

Thilak Rao

@Thilak

5 months

@levelsio How it started, how it’s going.

1

0

32

Thilak Rao

@Thilak

5 months

Double Siri's IQ — just add Meta Llama 3! (needs @private_llm to run locally)

4

28

Thilak Rao

@Thilak

4 months

@NickADobos I don’t steal or lie. We openly credit and often contribute code back to OSS projects that we use. Even Llama 3 can see through this grifter’s BS, why can’t you? This has turned into a proxy IQ test. Lmao

1

27

Thilak Rao

@Thilak

7 years

@aroraharshita33 @Nimbix I have known Vivek Na for years, and I am extremely disappointed to see you try to defame him for outing your lies and deceit. You do realise defaming someone is a criminal offense in India?

0

1

22

Thilak Rao

@Thilak

5 months

@tdinh_me This is why I don't run random shit from Gumroad or Setapp on my Mac.

3

0

22

Thilak Rao

@Thilak

5 months

Who needs the cloud? Llama 3 8B by @AIatMeta is cranking out 8.33 tokens/s on an M2 iPad Pro. All local, all offline. Download it from

1

4

24

Thilak Rao

@Thilak

5 months

@gjpx_ @levelsio Amen. Real therapy isn’t cheap everywhere, but it’s also freely available in many countries (In NL for instance, you can get it with a GP’s referral). Tragic that some vulnerable people who need it the most might fall prey to $9.99/month grifts like this instead.

1

0

22

Thilak Rao

@Thilak

5 months

@_Borriss_ Here's Llama 3 8B running fully locally and offline on my iPhone. Not a demo—you can buy this on the App Store today.

Private LLM: Secure, Local & Private AI Chatbot for iOS & macOS

Unlock the full potential of AI with Private LLM on your Apple devices. Enjoy local LLM capabilities, complete privacy, and creative ideation—all offline and on-device. Perfect for brainstorming,...

privatellm.app

Thilak Rao

@Thilak

5 months

Double Siri's IQ — just add Meta Llama 3! (needs @private_llm to run locally)

4

28

0

2

23

Thilak Rao

@Thilak

6 months

@paulg The EU's love for tech red tape is so grand, they've turned our web experience into a 'click to dismiss' marathon.

0

1

22

Thilak Rao

@Thilak

4 months

Hold onto your hats, folks! @private_llm just dropped a bombshell – the Meta Llama 3 70B model is now available for your Apple Silicon Macs with 48GB+ RAM

Private LLM

@private_llm

4 months

🖥️ v1.8.6 of Private LLM for macOS was released! 🎉 This release adds a 4-bit OmniQuant quantized version of the SOTA Llama 3 Instruct 70B🦙model by @AIatMeta . Works on all Apple Silicon Macs with 48GB or more RAM.

6

65

3

20

Thilak Rao

@Thilak

4 months

@KolinKoehl Good for you! If calling out a huckster is envy in your books, perhaps it’s time to reflect on your values.

0

21

Thilak Rao

@Thilak

5 months

@_glnarayanan @levelsio Lick less and build more.

1

19

Thilak Rao

@Thilak

5 months

@LewisNWatson @levelsio To paraphrase the memorable @JeffBezos line: His grifty SaaS MRR is my on-device subscription-free opportunity.

1

0

19

Thilak Rao

@Thilak

4 months

Llama 3 running locally on Apple Vision Pro, and it's out of this world! @private_llm

1

2

19

Thilak Rao

@Thilak

5 months

@soultraveler @levelsio @OpenAI @AnthropicAI Custom training? WTAF? 🤣 You’re so funny! AFAIK, it’s just Llama3-70B on @GroqInc with a custom system prompt. And before that it was Mixtral 8x7B on Groq with a custom system prompt.

0

16

Thilak Rao

@Thilak

6 months

@bindureddy Speaking of beating benchmarks, @private_llm users are already singing praises about Mixtral running locally on their macOS, replacing GPT 3.5 for them. The future of local LLMs is not just coming; it's here!

1

2

14

Thilak Rao

@Thilak

5 months

It’s here! Private LLM v1.8.1 for iOS just dropped! Grab the new Phi-3-mini-4k-instruct model from @Microsoft . It’s sleek, 4-bit OmniQuant, and supports 4k context. Runs fully on-device. Download now on any iPhone or iPad with 6GB+ RAM!

Private LLM

@private_llm

5 months

Private LLM v1.8.1 for iOS was just released. We’ve added support for downloading the amazing new Phi-3-mini-4k-instruct model from @Microsoft . The model is 4-bit OmniQuant quantized & supports 4k context length. It’s downloadable on all your i-devices with 6GB or more RAM. 🎉

4

46

3

15

Thilak Rao

@Thilak

6 months

@Jason Absolutely agree! Moving away from @vercel to self-host all our services has been a game-changer for @private_llm . We managed to bring our monthly costs down from $50 to just $20. When you’re bootstrapping from your bedroom, every dollar saved makes a huge difference.

2

0

14

Thilak Rao

@Thilak

5 months

You saw it here first, folks! One of @levelsio ’s woke minions trying to get me cancelled.

3

0

13

Thilak Rao

@Thilak

4 months

New swindle from @levelsio just dropped. AI and hard-tech startups in the EU are thriving. Even startups that had considered moving to the US are now reconsidering, in light of the recent US federal capital gains tax hike proposal. But our AI wrapper boi can’t sell shitty

4

0

12

Thilak Rao

@Thilak

4 months

@yar_vol @mattshumer_ The fine tune (dreambooth) is on user uploaded photos. If you think he’s fine tuning stock SD models, you’re mistaken. IIRC, he’d made a similar tweet like the one in question, taking credit for someone else’s work, in the early days of the photo thing. It’s a recurring pattern

0

12

Thilak Rao

@Thilak

4 months

@drusepth Please look up the term “implication” in a dictionary. There’s nothing needed to get it running in your browser, besides visiting the Web LLM page I’d linked to earlier. Thanks to the work of the Web LLM team, it’s been possible ever since Llama 3 was released, and Llama 2 has

1

0

11

Thilak Rao

@Thilak

6 months

My tweet got picked up by @palkisu on @firstpost , my views invite a pivotal inquiry: 'Cui bono?' in the regulatory debate on AI. Big tech firms seek to use regulation as a barrier against competition, while decels mask their support for these measures with the pretext of

2

0

11

Thilak Rao

@Thilak

5 months

This is so true! Some of the finest coders, product people, and designers are here posting on this app. Hard-working indie hackers who look up to some marketing guy LARPing as a startup guy instead of really talented people (I don't claim to be one of them) should really reflect

PEZO - Zoltán Pető

@ZoliPeto

5 months

@ddunderfelt @raspbfox @Thilak He is motivated by money, totally ignores the bad effect. He does not consider it his business. His business is making money. That’s why he weekly updates his lame “mrr” “/month” based bio, always picking the one which looks better. He is more of a marketing guy then a builder.

3

0

5

2

0

11

Thilak Rao

@Thilak

5 months

@Nikolay_S @tdinh_me Mykola, why isn’t your app sandboxed? Apple built the entire sandboxing technology in macOS to protect users from malware, and yet your app and apps you rent out on SetApp aren’t sandboxed. IIUC, the entire raison d'être of SetApp is to host apps that aren’t good enough pass the

4

0

11

Thilak Rao

@Thilak

5 months

Now available on the App Store

Private LLM

@private_llm

5 months

Private LLM for iOS v1.7.8 is now live on the App Store. 🎉 Experience the power of the latest Llama 3 8B Instruct model from @AIatMeta , running privately, fully on-device with no internet connection or telemetry. Works on all Pro, Pro Max iPhones and Apple Silicon iPads. Also,

10

23

165

0

1

9

Thilak Rao

@Thilak

1 year

Hey @friedberg , @Jason , & @theallinpod , remember your prediction on Ep. 131 about LLMs running on small machines? Meet @PersonalGPT , our on-device chatbot for iOS and macOS. Pre-orders are open now, with launch coinciding with #WWDC23 tomorrow. We're taking AI convenience and

Private LLM

@private_llm

1 year

Look ma, no Internet!

8

37

1K

2

8

9

Thilak Rao

@Thilak

6 months

@bindureddy It failed my Sally test prompt (unlike the 4-bit quantized Mixtral Instruct run locally via @private_llm )

0

1

9

Thilak Rao

@Thilak

6 years

@stuartfrisby It’s a fine balance between flexibility and constraints. That balance must never be lost. Design system should be continuously evolving and incorporating better ideas. But some design systems feel like they were drafted on a stone tablet.

2

1

9

Thilak Rao

@Thilak

5 months

@abnvx17 @levelsio Bhai Abhinav, have you ever heard the expression: सत्यमेव जयते? That’s true when it comes to building products as well. Those who indulge in मिथ्या (lies in Sanskrit), like this guy, have to work overtime shilling their fleece ware products. Those who build good products don’t

0

1

7

Thilak Rao

@Thilak

4 months

@bindureddy I've been playing with Llama 3 Smaug-8B all day on @private_llm . It's a solid Llama 3 finetune. Great job!

0

1

7

Thilak Rao

@Thilak

5 months

@petergyang @ollama I run it locally on my iPhone with @private_llm

Thilak Rao

@Thilak

5 months

Double Siri's IQ — just add Meta Llama 3! (needs @private_llm to run locally)

4

28

0

8

Thilak Rao

@Thilak

5 months

@paulg What’s the ratio of AI to non-AI applications for YC’s summer batch?

2

0

8

Thilak Rao

@Thilak

8 months

@rauchg Built @private_llm , an offline privacy focused AI chatbot for iOS and macOS. It has grown while exceeding our expectations.

0

3

Thilak Rao

@Thilak

9 years

Unicasts - Learn How To Build Ruby on Rails Web Apps From Design To Code

0

8

Thilak Rao

@Thilak

5 months

Exciting news! Our biggest iOS update ever for @private_llm is here, with an extensive range of new models from Google Gemma to Phi-2 and beyond!

Private LLM

@private_llm

5 months

10 days ago, we quietly released the biggest update to the iOS version of Private LLM since we launched 9 months ago, with over a dozen downloadable models, late last week. 🧶 on downloadable models in Private LLM for iOS, ordered by model size:

3

33

0

8

Thilak Rao

@Thilak

5 months

@GergelyOrosz Hopefully, future Agile gurus will stumble across this tweet when they Google "Spotify model" to emulate Spotify.

0

7

Thilak Rao

@Thilak

4 months

@CoderRadioShow @jupitersignal Hey @dominucco , and @ChrisLAS ! Thanks so much for the shoutout and trying out @private_llm with Llama 3 on the show. Really appreciate you sharing your experience with the app and these small but mighty models. We've also added support for Phi-3 Mini We’ll be bringing more

0

2

7

Thilak Rao

@Thilak

5 months

@localghost Achieved 15.56 tokens/sec on my iPhone using Google Gemma 1.1 2B, running entirely local and on-device with @private_llm . No laptop (or Internet) needed. Probably works much faster on my Mac

0

2

7

Thilak Rao

@Thilak

5 months

@VerdiK @Airbnb It’s because travel websites are notoriously bad at optimizing for web performance. In fact, @AirbnbEng has the best core web vitals score compared to their competitors. @nicj presented some stats around this in his Modern Metrics talk.

Modern Metrics | Nic Jansma | performance.now() 2022

What is a “modern” metric anyway? An exploration on how to measure and evaluate popular (and experimental) web performance metrics, and how they affect user ...

www.youtube.com

1

0

7

Thilak Rao

@Thilak

3 months

@bindureddy I'm doing my homework just fine 🤣

Private LLM

@private_llm

3 months

Sadly, not an excuse for programmers like us who run Llama 3 70B offline on their laptops. 😞

1

2

4

0

1

7

Thilak Rao

@Thilak

5 months

@liltechnomancer I'm glad people are calling him out on his nonsense. I did my part too.

Thilak Rao

@Thilak

5 months

I’m all for AI. I built AI products myself (and these aren’t API wrappers). But next token predicticting LLMs (aka “AI”) are no substitute for a trained therapist. There are multiple cases of people who need human help but instead went with an AI chatbot and were driven to

16

11

237

0

7

Thilak Rao

@Thilak

9 years

WebGL Fundamentals. This is Gold! http://t.co/BX5PW1vXIR

0

8

6

Thilak Rao

@Thilak

10 years

Why weren't you Interested in an executive position at Twitter? #AskBizStone

0

1

7

Thilak Rao

@Thilak

1 year

As our app @private_llm continues to get better, the reviews have been overwhelmingly positive! Just shipped v1.3, now boasting a 4.8/5 on the App Store. Thanks for making this indie dev journey so rewarding! 🎉 #AI #LLM #ChatBot

1

2

7

Thilak Rao

@Thilak

6 months

@ossia I've had my fair share of dealing with such people on the internet as a mod of @private_llm 's discord. I no longer view it as weird or rude; instead, I just wish them well on their mental health journey!

1

0

6

Thilak Rao

@Thilak

3 months

@GergelyOrosz @CraftConf Once the answer is 'yes, yes, yes' and you are 'default alive,' you don’t care about VC funding 🤣

2

0

7

Thilak Rao

@Thilak

6 months

@JsonBasedman Imagine the family gatherings: 'So, who's winning the GPU race this year?'

1

0

6

Thilak Rao

@Thilak

5 months

Excited to see @private_llm featured on @iMore today! Why wait for iOS 18? Experience cutting-edge open-source LLMs today by running them locally and fully on-device on your iPhone or Mac!

iMore

@iMore

5 months

Apple's major iPhone AI upgrade could have no cloud processing — initial wave of iOS 18 features "will work entirely on device"

1

7

0

2

6

Thilak Rao

@Thilak

10 years

A #Grunt Plugin for Testing under a Throttled Connection - http://t.co/BpT2SgyPKc #NodeJS

0

6

Thilak Rao

@Thilak

4 months

When your AI starts sounding like it's straight out of Lagos rather than Silicon Valley

Tony Zador

@TonyZador

5 months

ChatGPT overuses certain words, notably "delve" Why? Turns out it was trained by RLHF workers from Africa In Nigeria, “delve” is much more frequently used in business English than in England or the US. So we are ending up with an AI system that writes slightly like an African

15

103

366

0

6

Thilak Rao

@Thilak

8 months

@agreeahmed @erikgoinsHQ @ycombinator Classic YC approach. Start by creating a video for a concept that’s not yet developed, then pivot repeatedly—testing various ideas until one resonates with the market or the budget for advertising dries up. Probably if Airbnb were to apply for YC today, they’d be rejected

2

0

6

Thilak Rao

@Thilak

5 months

@localghost Getting 5.5 tokens per second on the iPhone 15. It's not a bad trade-off considering you can now have a GPT 3.5-grade LLM in your pocket without a Mac or the internet.

Thilak Rao

@Thilak

5 months

Just got Llama 3 8B Instruct by @Meta up and running on my iPhone with @private_llm , running fully on-device on my iPhone with a full 8K context on 8GB devices. Coming to all iPhones with 6GB or more RAM this weekend.

10

35

275

1

6

Thilak Rao

@Thilak

4 months

Say hello to the new Phi-2 based Dutch LLM!

Bram

@BramVanroy

4 months

👱‍♀️ Daar is ze dan, there she is! Fietje has arrived, a small and powerful #LLM for #Dutch . 🇧🇪🇳🇱 Fietje, based on @MSFTResearch @phi2 , is 2.5x smaller than models like GEITje 7B Ultra, but manages to match their performance in benchmarks. 🚀 Thread 👇

2

6

27

0

6

Thilak Rao

@Thilak

4 months

@ShaulGoo @mikb0b @mattshumer_ Well then, you are not a part of his core audience. His core audience is the non/semi tech audience who thinks he single handedly builds AIs, whatever that means. That’s the image he’s tries to cultivate with blatant plagiarism.

0

6

Thilak Rao

@Thilak

5 months

@minchoi You missed Llama 3 8B running locally on iPhone (no Mac, no internet)

Private LLM

@private_llm

5 months

Private LLM for iOS v1.7.8 is now live on the App Store. 🎉 Experience the power of the latest Llama 3 8B Instruct model from @AIatMeta , running privately, fully on-device with no internet connection or telemetry. Works on all Pro, Pro Max iPhones and Apple Silicon iPads. Also,

10

23

165

0

6

Thilak Rao

@Thilak

1 year

As we gear up for this year's #WWDC23 , I'm thrilled to share a project my brother, Jeethu Rao, has been working on - Personal GPT. It's a chatbot for iPhones that operates entirely on-device, a first of its kind. This isn't your typical AI assistant. It runs offline, making

Jeethu Rao

@jeethu

1 year

Time to finally reveal what I’ve been building, lately. Fully private and on-device. Pre-order open on the App Store at an early adopter price (read: throwaway price). Releasing 5th June. DM me the receipts if you can’t wait and want the beta running on your phone, RN!

2

5

13

0

1

5

Thilak Rao

@Thilak

5 months

RIP Devin, which was all the hype recently. These guys have replicated it with MIT-licensed code.

carlos

@_carlosejimenez

5 months

SWE-Agent is an open-source software engineering agent with a 12.3% resolve rate on SWE-Bench! Check out SWE-agent in action at Repo:

30

113

561

0

6

Thilak Rao

@Thilak

8 months

Exciting news for macOS users! Private LLM v1.7.0 now includes two powerful 7B models: openchat-3.5-0106 and CodeNinja-1.0-OpenChat-7B. Plus, a much-awaited feature: customizable system prompts.

Private LLM

@private_llm

8 months

Private LLM for macOS v1.7.0 has just been released. We've added support for two 7B models: the new SOTA openchat-3.5-0106 model by @openchatdev and excellent coding model CodeNinja-1.0-OpenChat-7B by @beowolx . Also, a frequently requested feature: customizable system prompts.

2

13

0

5

Thilak Rao

@Thilak

4 months

@mikb0b @mattshumer_ > Apache actually only requires attribution within source code releases, not anywhere else (*not a lawyer) Please learn to read. I assure you it’s easy, and might save you a lot of legal hassles in the future.

2

0

6

Thilak Rao

@Thilak

1 year

Big news! Personal GPT v1.1 is here with a complete overhaul of our iOS codebase. Experience the speed with our new Metal backend that's 1.5x faster! And with improved conversational memory, your chats just got a lot more engaging. #AI #GenerativeAI #privacy

Private LLM

@private_llm

1 year

🚀 We've supercharged Personal GPT with the new v1.1 update! Here's what's new 🤘 A faster, more efficient Metal backend ⚖️ Reduced app size 🗣️ Much improved conversations, for more engaging and natural interactions 👨‍💻 Smarter syntax highlighting for code

0

4

6

0

1

5

Thilak Rao

@Thilak

9 years

After listening to @wycats and looking at Angular 2.0 vs Aurelia, I think it's about time I invest in #EmberJS #jschannelconf

1

5

Thilak Rao

@Thilak

5 months

@nielsthebuilder More seriously: State of the art model quantization tech (OmniQuant) and Apple’s Metal for running LLM inference on iPhone’s GPU.

0

4

Thilak Rao

@Thilak

4 months

@ZoliPeto Grifters be grifting, I guess?

1

0

5

Thilak Rao

@Thilak

5 months

@Jason Who was the photographer?

0

5

Thilak Rao

@Thilak

5 months

@minchoi Llama 3 🦙❤️ Siri

Thilak Rao

@Thilak

5 months

Double Siri's IQ — just add Meta Llama 3! (needs @private_llm to run locally)

4

28

0

5

Thilak Rao

@Thilak

10 years

Save Objects and Arrays, Numbers or Other Data Types into localStorage with Lockr http://t.co/ifK0G9qZg1 #WebDev #JavaScript

0

3

Thilak Rao

@Thilak

5 months

It's surprising that a 3-year-old iPhone SE can support 4K token context with @h2oai Danube Chat 1.8B

Private LLM

@private_llm

5 months

The new H2O Danube Chat 1.8B model in Private LLM from @h2oai runs very well on even a 4 year old iPhone SE2 with only 3GB of RAM. 4k context length, great for building you own on-device RAG Shortcuts!

1

2

16

0

5

Thilak Rao

@Thilak

5 months

@localghost Absolutely thrilled to run llama 3 8B on my iPhone with @private_llm ! But, Apple, let’s not stop there – give us background GPU execution!!

1

0

5

Thilak Rao

@Thilak

6 months

@manoj_ahi Built @private_llm - the AI chatbot that stays private like your diary and works offline like your old-school Gameboy. Who needs the internet anyway?

0

4

Thilak Rao

@Thilak

5 months

@enjoypolosfu @Meta @private_llm @GroqInc Comparing apples to apples

Jeethu Rao

@jeethu

5 months

@emollick NVIDIA claims an H200 can do 3000 tokens/second.

2

0

15

1

0

4

Thilak Rao

@Thilak

5 months

@soultraveler @levelsio @OpenAI @AnthropicAI @GroqInc the photo thing is dreambooth which involves some training, the interior one doesn’t. diffusion models are cribbed from civitai. the applicant thing is also just prompts with a php and jQuery crud

1

0

5

Thilak Rao

@Thilak

5 months

@helloitsolly Brace for his shills and sock puppet accounts to attack you.

0

5

Thilak Rao

@Thilak

10 years

7 Things I Wish Every Search Box Did by @dessaigne http://t.co/G2JTiXcctY #WebDev http://t.co/oXFIJqDnUS

0

5

Thilak Rao

@Thilak

4 months

@vrilkroket @levelsio Dear racist anon anti-vaxxer pig, I love this country and I pay taxes here. Your huckster idol @levelsio took advantage of the social security net and free education provided by the Netherlands and the European Union and is now eating rice in some South East Asian country

1

0

5

Thilak Rao

@Thilak

4 months

@mikb0b @mattshumer_ Yep. Screenshots are not derived works, but modified source and compiled binaries are. Regardless of the ethics and morality of it, posting a cropped video of someone else’s website and passing it off as one’s own is very cringe, and reeks of neediness. I’ve said this before, the

1

0

5

Thilak Rao

@Thilak

7 months

Exciting sneak peek: Google Gemma 2B coming soon on @private_llm

0

5

Thilak Rao

@Thilak

1 year

After months of tireless work and dedication, the day has finally come. I’m incredibly proud of @jeethu for making @PersonalGPT a reality. Check it out now on the App Store! #WWDC23 #LaunchDay

Private LLM

@private_llm

1 year

🎉 It's launch day! Personal GPT is now live on the App Store, coinciding with #WWDC23 Experience true privacy with our offline support. Explore new horizons with your private AI chatbot on all Apple devices. #GenerativeAI #Privacy

0

3

0

2

Thilak Rao

@Thilak

10 years

10 Most Common #JavaScript Mistakes http://t.co/6FpcKUDj98

0

1

2

Thilak Rao

@Thilak

4 months

This eu/acc thing is quickly turning into an IQ test. The symbol to look out for to stay away from midwits. Because they fell for such a simple con. 😂

0

4

Thilak Rao

@Thilak

5 months

@mauriciogior @LewisNWatson @levelsio @JeffBezos Yes, and there’s also a bigger macOS version with even bigger models (free download for anyone who has the iOS app from the Mac App Store) battery perf depends on the device. It’s pretty good on the latest pro and pro max devices

1

0

4

Thilak Rao

@Thilak

4 months

Apple: "We care about your privacy." Also Apple: approves apps with more red flags than a Soviet parade 🚩🚩🚩

Private LLM

@private_llm

4 months

An app’s App Store privacy nutrition labels vs the same app’s privacy policy. And the privacy policy is just the tip of the iceberg. 😅 Many such cases. “We value your privacy…”.

0

3

13

0

4