Guido Appenzeller @appenz Twitter profile

Last Seen Profiles

@Rain_NFT

@EPutaski35751

@CharuwanPu30846

@yxpiee

@colinshenzhen

@bokeplokalmalam

@emetswife

@kw_201517

@Ulyth

@ba_nran

@Philotes66

@MayorofLondon

@igobykenn

@SNDS_SN

@BinorRaja

@bilal__jamal

@sakumaru___39

@toramaru0032

@dvisser2001

@Oppai_Panya

@softskillpunk

@HwayLqhab

@natalia_shev4uk

@salbbsna1

@Alfoncsyons1

@jandakembangstw

@ChocolateThoug2

@AcapulcoShore

@ba_nran

@respawnador

@sixfiveotwo

@POLKProfLearn

@NERDYXNiNJA

@gunchan_1130

@bokeplokalmalam

@Teeda4real

Guido Appenzeller

@appenz

2 years

1/7 Spent the week-end with ControlNet, a new approach to have precise, fine-grained control over image generation with diffusion models. It's huge step forward and will change a number of industries. Here is an example.

42

278

2K

Guido Appenzeller

@appenz

6 months

About 15% of all Stanford undergrads (of all majors) are learning how to build LLMs. Stats for @chrmanning 's CS224N Natural Language Processing with Deep Learning below. That makes sense. LLM's are becoming a basic systems component like compute, networking and storage.

17

66

334

Guido Appenzeller

@appenz

1 year

🔥 In a recent post Sequoia's @DavidCahn6 argues AI infra is overbuilt: - NVIDIA GPU revenue is $50b/y - This requires $200b in "AI revenue" - There is only $75b in "AI revenue" Thus there is a $125b hole I strongly disagree. AI Infra will be huge. Grab🍿and read on (🧵1/7).

12

62

318

Guido Appenzeller

@appenz

2 years

6/7 If you want to try this out, it's realtively easy to get running if you have a Windows PC with a high-end Graphics Card (12 GB RAM is ideal, but less works too). I used the Automatic1111 extension here.

GitHub - Mikubill/sd-webui-controlnet: WebUI extension for ControlNet

WebUI extension for ControlNet. Contribute to Mikubill/sd-webui-controlnet development by creating an account on GitHub.

github.com

3

20

303

Guido Appenzeller

@appenz

4 years

I am excited to announce that I am joining @Intel as CTO of the Data Platforms Group. Intel is not without challenges. But I think strength in CPUs, a diverse portfolio, massive scale and Navin’s great team will make it successful. And it's awesome to work with @PGelsinger again!

39

25

290

Guido Appenzeller

@appenz

1 year

1/5 For companies doing Generative AI, finding enough GPUs is a difficult and expensive. We’ve seen companies spend 80%+ of total capital raised on compute resources. To help them @casado , @BornsteinMatt and I wrote a post how to navigate the cost of AI.

11

69

285

Guido Appenzeller

@appenz

2 months

1/6 I fine-tuned a @bfl_ml 's Flux.1-dev LoRA on myself. It fine tunes really well, easier than SDXL although not quite as easy as SD 1.5. Tuning was done on @replicate , the model hosted on @huggingface . Total cost is about $7 for 75 minutes on an A100. Full instructions below.

10

25

283

Guido Appenzeller

@appenz

2 years

2/7 The basic idea of control net is that your diffusion model works in tandem with a second model trained on a specific task. This is the sd15_scribble model. It helps Stable Diffusion 1.5 to interpret sketches, and turns my "sketch" of an Owl into an awesome drawing.

3

10

164

Guido Appenzeller

@appenz

2 years

7/7 Congrats to Lvmin Zhang and Maneesh Agrawala @magrawala at Stanford for developing this technology, the original research publication is here.

Adding Conditional Control to Text-to-Image Diffusion Models

We present ControlNet, a neural network architecture to add spatial conditioning controls to large, pretrained text-to-image diffusion models. ControlNet locks the production-ready large diffusion...

arxiv.org

4

9

152

Guido Appenzeller

@appenz

2 months

1/3 New benchmarks from @MLPerf , and they include the first good B200 numbers that I have seen. 11,264 tokens/s for Llama 2 70b is crazy, and about 3.7x the performance of the H100🤯 The bar for every AI silicon start-up out there just went up. Some thoughts below.

10

29

146

Guido Appenzeller

@appenz

2 years

1/8 I created a picture of three robots posing to spell the letters "AI". A month ago creating this would have taken me days, this week-end it took me 15 minutes. Below an overview of ControlNet and OpenPose. Safe to say this will change image generation forever.

2

21

139

Guido Appenzeller

@appenz

1 year

This "investor" currently has the #1 and #5 spot on GitHub's trending repos. I love working with this team.

3

11

117

Guido Appenzeller

@appenz

2 years

5/7 And last but not least, ControlNet combines with fine-tuning via DreamBooth. I can fine-tune a model on myself (like in the James Bond picture), and then use it to render myself into a specific scene. Here is our daughter who wanted to be Black Widow.

12

6

119

Guido Appenzeller

@appenz

2 years

1/5 Meta launched their GPT-3 competitor LLaMA today. Here is a quick analysis of how it stacks up, how open it is and how it changes the industry landscape.

LLaMA: Open and Efficient Foundation Language Models - Meta Research

We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to...

research.facebook.com

3

31

110

Guido Appenzeller

@appenz

6 years

I am super excited about my next adventure and working with the amazing team at @Yubico . Wohoo!

Yubico | #YubiKey

@Yubico

6 years

Meet the newest addition to our executive team: Guido Appenzeller! Guido joins us as the Chief Product Officer of Yubico to focus on product development and strategy. Learn more about Guido in our latest interview Q&A:

0

4

20

18

3

104

Guido Appenzeller

@appenz

2 years

And apparently you can also use it to control lighting in a 2D photo.

6

4

96

Guido Appenzeller

@appenz

3 years

Exciting additions to Intel's executive team. A huge welcome to my former colleague from @vmware Greg Lavender joining as CTO and my Ph.D. advisor Nick McKeown stepping up to GM. And thanks to Navin Shenoy for dragging me to @intel , I will miss you!

Intel Makes Organizational Changes to Strengthen Execution, Innovation in Critical Business Areas

Intel announces the addition of two new technology leaders to its executive leadership team, as well as several changes to Intel business units.

www.businesswire.com

1

8

94

Guido Appenzeller

@appenz

1 year

1/3 The amazing team at @lmsysorg have their new leaderboard up, and MPT-30B-Chat is the new #1 for open-source models. Congratulations @MosaicML ! Technically the model they tested is not fully OSS (it is CC-BY-NC-SA-4.0). But this doesn't matter.

Chatbot Arena Leaderboard Week 8: Introducing MT-Bench and Vicuna-33B | LMSYS Org

In this blog post, we share the latest update on Chatbot Arena leaderboard, which now includes more open models and three metrics: Ch...

lmsys.org

2

26

93

Guido Appenzeller

@appenz

2 years

4/7 ControlNet can also work with Pose. Here is a super interesting demo of taking the pose from a 3D character editing tool (Houdini) and using it to generate an exact pose in an image.

From the StableDiffusion community on Reddit: I'm working on API for the A1111 ControlNet extensi...

Explore this post and more from the StableDiffusion community

www.reddit.com

1

3

86

Guido Appenzeller

@appenz

9 months

1/8 Copilot for @Microsoft @Office came out today. As a reformed CTO and fan of @GitHub Copilot, I love the idea and gave it a spin. TL;DR: It's a mess and not anywhere close to adding value. A thread🧵.

8

10

87

Guido Appenzeller

@appenz

6 years

I am leaving @VMware . It's been four amazing years working with a great team of people. Thanks to all of you! Blog post here:

31

7

85

Guido Appenzeller

@appenz

4 years

Flew to Catalina last week, and suddenly I had this feeling I was staring at my macOS desktop.

3

2

85

Guido Appenzeller

@appenz

3 years

If you combine @Intel and @VMware , you get a great home server. New blog post on installing ESXi on an Intel NUC11 with a custom install image (as ESXi still lacks drivers).

4

20

84

Guido Appenzeller

@appenz

1 year

Understanding technology and market implications of Generative AI and Foundation Models is critical for navigating the tech economy. Based on what we read at @a16z to get up to speed, @derrickharris @BornsteinMatt and I put together a reading list.

1

28

79

Guido Appenzeller

@appenz

2 years

1/6 @gptzero soft launched, and this is an opportunity to look at the state of the art in detecting AI generated text. It works surprisingly well. For a simple prompt, it in most cases identifies AI generated parts. However with a good prompt you can fool it.

5

13

72

Guido Appenzeller

@appenz

3 years

Naru and myself announcing Mt Evans, @intel 's Infrastructure Processing Unit. Make no mistake, this will change how data centers will be architected for both hyperscale and enterprise alike. CPUs are for tenants, CSP code runs on the IPU.

IPU and Mt. Evans - Architecture Day 2021 | Intel Technology

We’re in the middle of an infrastructure revolution, and at Architecture Day we unveiled our IPU, or Infrastructure Processing Unit. Oak Springs Canyon and A...

www.youtube.com

2

21

71

Guido Appenzeller

@appenz

8 months

Goody-2, the most responsible AI model. Brilliant 😂

4

10

71

Guido Appenzeller

@appenz

5 years

Ruby programmers are more polite than Python programmers according to Stanford Linguistics @jurafsky . Anecdotal evidence seems to confirm.

5

24

69

Guido Appenzeller

@appenz

2 years

Thread: I trained Stable Diffusion on photos of myself, and can generate pictures of myself purely based on text input. So for example entering "A photo of <appenz> as Captain America" will generate this. Total cost is about $0.30 of Google TPU time and $0.001 per picture.

6

11

64

Guido Appenzeller

@appenz

6 years

VMware is acquiring @Heptio . We are very, very serious about containers. Welcome on board @cmcluck , @jbeda and the entire the Heptio team!

0

21

62

Guido Appenzeller

@appenz

7 years

We have a new GM of VMware’s Cloud Services. And @virtualmilin is the best person I could imagine for this job. Congratulations Milin!

10

9

61

Guido Appenzeller

@appenz

6 years

Wohoo! After 20 hours in the air we made it from California to Punta Cana, Dominican Republic in a single engine propeller plane. An awesome way to end the year.

11

1

59

Guido Appenzeller

@appenz

2 years

@dweekly My kids have started using GPT-3 for school assignments, quoting it as a source. It makes me wonder if teaching simple knowledge questions (e.g. "Define acid rain.") is worth it today, or if it is the equivalent of dividing large numbers manually without a pocket calculator.

5

4

57

Guido Appenzeller

@appenz

1 year

7/7 In summary, there is no hole. AI will be a ubiquitous component in any product that contains software. A $50b spend on GPUs infra can easily be amortized over a $5T worldwide IT spend. And I am super excited to invest in AI infra and the applications that are powered by it!

4

5

56

Guido Appenzeller

@appenz

6 months

Super excited that @a16z raised a new infra fund. But I am even more excited to be part of the amazing team that will invest this fund.

martin_casado

@martin_casado

6 months

We've raised a $1.25B infrastructure fund! We love all infra, compute, network, storage, databases, data science, gen AI, dev tools ... from silicon to UIs. Infra is the true root of value in tech. And we're deepening our commitment to it.

37

39

526

2

46

Guido Appenzeller

@appenz

2 years

3/7 ControlNet has a number of these models and a framework to train new ones. The model used for the Bond photo uses the output of the Canny edge detector and creates an image with very similar edges. This creates a very close match in terms of overall layout.

1

55

Guido Appenzeller

@appenz

2 years

I wrote a ChatGPT tool. It watches the clipboard on macOS. If it finds a trigger (default "@@"), it replaces the Clipboard with its ChatGPT output. 1. Copy "@ @Capital of France?" 3. wait 1-2 seconds 4. Paste "Paris..." Super useful, 70 lines of Python.

GitHub - appenz/macLLM: Tool to integrate LLM's like GPT-3 with macOS

Tool to integrate LLM's like GPT-3 with macOS. Contribute to appenz/macLLM development by creating an account on GitHub.

github.com

1

14

55

Guido Appenzeller

@appenz

5 years

Yubico Authenticator App for iOS Now Supports NFC. Thanks to everyone who helped with beta testing!

2

20

53

Guido Appenzeller

@appenz

6 years

How hard is it to fly a small plane from the SF Bay Area to the Carribean? Well, I am in the process of finding out. If you want to follow the adventure, I am posting updates on my home page.

11

0

55

Guido Appenzeller

@appenz

4 years

Blog Post: I recently updated my home office to become more centered around video conferencing. Here are some lessons learned, the gear I use and why we are all just streamers.

1

5

54

Guido Appenzeller

@appenz

1 year

A new image model just launched, and it's a very good one. Congrats to the team at @ideogram_ai !

Ideogram

@ideogram_ai

1 year

We're excited to announce the formation of Ideogram AI today! Our mission is to help people become more creative through Generative AI.

70

141

630

2

20

38

Guido Appenzeller

@appenz

6 years

AWS Outpost - AWS is coming on premises and it is integrated with VMware. It’s hard to overstate how big this is. @PGelsinger @ajassy @vmwarecloud .

0

20

49

Guido Appenzeller

@appenz

7 years

First public tech preview of NSX for Azure. Great work Team NSX!

Britton Johnson 🎙️

@vcixnv

7 years

@appenz @scott_lowe Live Demoing NSX on AWS and Azure amazing stuff! @vmwarensx @VMworld

0

8

12

1

45

48

Guido Appenzeller

@appenz

1 year

5/7 But most importantly, it misses the magnitude of the AI revolution. AI Models are an infra component like CPUs, databases & networks. - Today almost all software uses CPU/DB/Network - In the future the same will be true for models So all software will be powered by AI infra.

4

9

47

Guido Appenzeller

@appenz

5 years

Congrats to @block_one_ team for an amazingly #b1june event to day here in Washington DC today. It was great to be on stage, we are excited about EOS using Yubikeys @Yubico and future development!

2

5

45

Guido Appenzeller

@appenz

9 months

Very sad to see IEEE turning on its own members like this. Their mission is to promote engineering research and innovation, and instead they give their front page to a political activist with zero engineering background who uses it to attack AI research.

martin_casado

@martin_casado

9 months

Sad to see @IEEESpectrum with a big fuck you to open source. It’s one thing to print irrelevant shit (which it’s been doing for years). It’s quite another to be actively anti tech. Good thing few people under the age of 50 read Spectrum any longer.

36

47

327

3

8

47

Guido Appenzeller

@appenz

6 years

SMS for two factor authentication or password reset is unsafe and will disappear. If you don't believe it, read this article about recent arrests of people who used SIM swapping to reset accounts and steal crypto currencies in the @mercurynews

Cryptocurrency thief cops to million-dollar hacking scheme as Santa Clara County tech squad builds...

Boston-area resident Joel Ortiz agrees to 10-year prison sentence for ‘SIM swap’ thefts of bitcoin and other digital funds, netting over a million dollars stolen from several victims la…

www.mercurynews.com

4

27

46

Guido Appenzeller

@appenz

1 year

1/4 ServiceNow & Huggingface just released StarCoder, a language model for code. The impressive stats are 15.5B parameters, trained on 1T tokens, 80+ programming languages. Here is a quick test how it stacks up against Co-pilot.

bigcode (BigCode)

huggingface.co

1

8

44

Guido Appenzeller

@appenz

3 years

Big day for @Intel as @PGelsinger and Amin Vahdat announce the collaboration on Mt Evan's. With heterogeneous compute becoming the norm, the IPU is the most strategic socket in a modern hyperscale data center and Intel+Google built a very strong product.

Susan Bobholz

@SBobholz

3 years

Amin Vahdat @Google talking about our partnership to create the Mount Evans IPU. #IntelON #IamIntel

0

5

11

0

8

44

Guido Appenzeller

@appenz

3 years

After three decades of air travel, this is my favorite international terminal building.

4

1

42

Guido Appenzeller

@appenz

2 months

LoRA stacking for @bfl_ml #Flux is out and it works for combining people into one photo. Some bleed of the LoRAs concepts, perfectionists should use inpainting. Below: - Both LoRA's stacked - My LoRA, different ages - Charlotte's LoRA Model:

5

6

41

Guido Appenzeller

@appenz

3 years

We are launching @Intel 's Infrastructure Processing Unit (IPU), a strategic pillar of our data center strategy. It may look like a SmartNIC, but main benefits are actually storage virtualization and workload provisioning for Clouds. #SixFiveSummit

Intel Unveils Infrastructure Processing Unit

Intel’s IPU improves data center efficiency and manageability and is the only one built in collaboration with hyperscale cloud partners.

www.intel.com

1

12

41

Guido Appenzeller

@appenz

2 months

Flux.1 dev with the right LoRA's is just ridicolously good for Watercolor style profile images. Via @venturetwins , she is working on a tutorial how to do it.

1

0

40

Guido Appenzeller

@appenz

1 year

1/5 @StabilityAI released their LLM today. It's called StableLM and they are planning versions up to 65b parameters. I tried the 7b parameter model and unfortunately it is not really usable yet. Example:

6

9

37

Guido Appenzeller

@appenz

7 years

Woo-Hoo! Pere Monclus is our new CTO of VMware's networking business unit. Looking forward to shaking up the networking industry together!

2

10

37

Guido Appenzeller

@appenz

1 year

4/7 Second, you don't need $1 of electricity for each $1 of GPU. An H100 PCIe costs ~$30k and uses ~350W of power, maybe 1kW with server/cooling. At $0.10/kWh, this H100 costs about $0.15 in electricity over a 5 year life cycle for each $1 spent on hw.

NVIDIA H100 Tensor Core GPU

Tap into exceptional performance, scalability, and security for every workload with the NVIDIA H100 Tensor Core GPU.

www.nvidia.com

2

37

Guido Appenzeller

@appenz

1 year

6/7 Networking infra spend is $200b+ per year. Does this create $800b in "networking software" revenue? No, but Google uses networking infra to sell ads and it shows up as ad revenue. Or Microsoft sells O365. The label for infra spend and revenue generated by it are different.

2

34

Guido Appenzeller

@appenz

5 years

Incredibly humbled that Nick McKeown, Isaac Keslassy & I were invited to write one of 20 editorials reflecting on 50 years of networking research for SIGCOMM. We have come a long way since we tried to explain why buffers are too large back in 2003.

Sizing Router Buffers (Redux)

Nick McKeown, Guido Appenzeller, Isaac Keslassy Abstract The queueing delay faced by a packet is arguably the largest source of uncertainty during its journey. It therefore seems crucial that we...

ccronline.sigcomm.org

0

7

34

Guido Appenzeller

@appenz

7 years

In a cloud+container world, we are getting pretty close to this. Networking is a software industry.

2

18

35

Guido Appenzeller

@appenz

2 years

In 2005 Isabelle and I climbed Kilimanjaro and this year I did it with our oldest kid. Not much has changed. The gear is a bit better, you now have 5G coverage at 19,341 ft altitude, Lemosho is now the favored route, but in the end it's all about an incredibly beautiful mountain.

2

0

35

Guido Appenzeller

@appenz

11 months

Germans (like me) talking about AI is is apparently against @OpenAI 's content policy. Americans are fine. This is why we need the open source AI. We don't want a few large companies dictate "AI ethics" for all of us.

8

3

32

Guido Appenzeller

@appenz

7 years

It’s official, @VeloCloud is joining the VMware family. A huge welcome to the entire team! @inetnut @ajitmayya @swoosnpx

VMware News

@vmwarenews

7 years

Breaking News: VMware announces intent to acquire VeloCloud Networks: by @shekar_ayyar

3

165

148

2

15

32

Guido Appenzeller

@appenz

2 years

@bilawalsidhu Yes, I saw it on Reddit (and upvoted it!). That looks really amazing. 100% agreed that this will change interior design (or for that matter most design). Is this based on depth? Or Canny??

1

0

32

Guido Appenzeller

@appenz

1 year

Asked GPT-4 to remove unnecessary boilerplate from a message from my bank ( @Chase ), and it reduced its length by 85%. This tells us a lot about the power of AI. And about Chase.

3

6

29

Guido Appenzeller

@appenz

1 year

1/3 New paper on Transformer Models with long context windows. This is an important result as it at least for now it means that LLM based search and knowledge extraction over large amounts of data will continue to use vector databases.

1

7

30

Guido Appenzeller

@appenz

2 years

If you wonder how AI based interior design may look like in the future, this is probably a glimpse of it.

Bilawal Sidhu

@bilawalsidhu

2 years

ControlNet experiment where I'm toggling through different styles of contemporary Indian décor, while keeping a consistent scene layout. Loving how ControlNet is putting the artists back in control of AI image generation process. 🧵Thread #ControlNet #StableDiffusion #EbSynth

41

285

2K

0

3

27

Guido Appenzeller

@appenz

1 year

1/2 Detection of AI generated text: 1. It discriminates against essays from non-native speakers 😠 2. AI can be used to make that human generated text *less* likely to be flagged. Just tell it to re-write your essays to be more "native"🤣

3

10

29

Guido Appenzeller

@appenz

8 years

OVH is buying vCloud Air. VMW is going back to its roots of being a Software/SaaS company.

0

28

Guido Appenzeller

@appenz

6 years

Customer round table with the @vmwarensx team at the Oktoberfest. This is doing business German style.

2

0

28

Guido Appenzeller

@appenz

2 years

@antona23 Yes, apparently this runs on M1 macs, Reddit thread below. Now it will be slow. You may want to consider running it on a Google Colab or something similar instead.

1

0

29

Guido Appenzeller

@appenz

5 years

We are super excited about the partnership between RSA and Yubico. RSA pioneered strong authentication with their tokens and now we are jointly delivering FIDO2 based solutions to the market. Looking forward to working with @Zulfikar_Ramzan , Jim Ducharme and the entire RSA Team!

RSA

@RSAsecurity

5 years

NEW from RSA: We're partnering with @Yubico ! Learn how we're tackling the growing challenge of #DigitalRisk together with a FIDO2-enabled device that delivers enterprise-grade security and risk-based #authentication :

0

10

21

3

6

28

Guido Appenzeller

@appenz

6 years

Excited about @_TomGillis joining VMware as our new GM for the Networking & Security BU. Welcome on board! #NSX

VMware Welcomes Tom Gillis as SVP & General Manager, Networking and Security Business Unit

Tom Gillis, the co-founder of Bracket Computing, will join VMware as senior vice president and general manager of the Networking + Security Business Unit

blogs.vmware.com

0

6

29

Guido Appenzeller

@appenz

2 years

My daughter asked GPT-4 for a summary of Shakespeare's Hamlet for Millennials. LLM's are amazing 😂

3

4

28

Guido Appenzeller

@appenz

2 years

Wohoo! Kubernetes is getting a native API gateway based on @envoyproxy . Very cool as it allows you to run service mesh @IstioMesh and API gateway on the same foundational technology. Congrats @mattklein123 @Tetrateio @VMware & the steering comittee!

Envoy API Gateway: The Gateway to a New Frontier

Today, the Envoy community announced an exciting new project: Envoy Gateway. The project unites industry leaders to streamline the benefits of application gateways powered by Envoy.

tetrate.io

1

8

27

Guido Appenzeller

@appenz

1 year

We at @a16z are launching a new program to support Open Source AI. More in the post from @BornsteinMatt .

Matt Bornstein

@BornsteinMatt

1 year

[New program] a16z Open Source AI Grants Hackers & independent devs are massively important to the AI ecosystem. We're starting a grant funding program so they can continue their work without pressure to generate financial returns.

70

265

1K

0

1

26

Guido Appenzeller

@appenz

4 years

Apparently "Guido" is a "bad word" and prohibited in @Ring 's support forums. Before you buy a Ring Door Bell, you may want to check if your legal name is in compliance with their policy. Otherwise, change your name first (or buy Nest instead).

4

2

26

Guido Appenzeller

@appenz

5 years

Practical security 101: Can I open a bottle with a Yubikey? Yes, you can, but please don't do this unless you have a backup key enrolled.

Yubico | #YubiKey

@Yubico

5 years

@vcsjones How's this? 🙂

2

8

1

6

23

Guido Appenzeller

@appenz

3 years

"Wouldn't you rather work from Hawaii?" @Intel 's Robert Noyce predicting 40 years ago that half of the work force could work remotely in the future, and that communication will be the gating element (I am looking at you, bad hotel Wifi) #agedlikefinewine

Robert Noyce

In this Classic Clip, Dr. Robert Noyce, Vice Chair of Intel and a pioneer in the electronics industry, talks about the future of technology in the workplace....

www.youtube.com

1

4

25

Guido Appenzeller

@appenz

2 months

2/6 I used @lucataco93 's trainer on Replicate yesterday, although Replicate came out with their own trainer later in the day. It looks like it is easier to use:

Fine-tune FLUX.1 with your own images

We've added fine-tuning (LoRA) support to FLUX.1 image generation models. You can train FLUX.1 on your own images with one line of code using Replicate's API.

replicate.com

1

4

24

Guido Appenzeller

@appenz

8 years

Congrats @MichaelDell and Joe @EMCcorp . A historic deal that is leading major change in the IT landscape!

0

19

24

Guido Appenzeller

@appenz

1 year

Great slide from Bill Dally at @hotchipsorg 2023 on NVIDIA's performance gains. This graph, parallelism (i.e. transformers) and optimization (e.g. flash attention) are the foundation for the Generative AI boom. It's all about infra. More here:

2

24

Guido Appenzeller

@appenz

2 years

ControlNet is really amazing at taking computer game characters and turning them into photos. This is Alexstrazsa from World of Warcraft. #worldofwarcraft #stablediffusion #ControlNet

1

5

24

Guido Appenzeller

@appenz

5 months

New podcast from @BornsteinMatt and myself on what is happening in Generative AI right now.

Open Models and Maturation: Assessing the Generative AI Market | Andreessen Horowitz

In this episode of the AI + a16z podcast, a16z partners Guido Appenzeller and Matt Bornstein discuss the state of the generative AI market, about 18 months after it really kicked into high gear with...

a16z.com

0

4

24

Guido Appenzeller

@appenz

4 months

Modern AI hardware is just wild: - Left: 1U 2xGH200 server, 2x 2.7kW power - Right: Cabinet spec for Equinix SV2 is 2.4-5kW power So a single server exceeds the power limit for the entire cabinet. Which has space for 42 servers 🤯 And Blackwell will make this even worse.

3

5

24

Guido Appenzeller

@appenz

7 months

Nice flex from the @NVIDIA PR team: Press release includes quotes from the CEOs of Amazon, Microsoft, Google Deepmind, Meta, Oracle, Tesla and OpenAI 😲 Picks & Shovels in an early gold rush are one hell of a business.

NVIDIA Blackwell Platform Arrives to Power a New Era of Computing

Powering a new era of computing, NVIDIA today announced that the NVIDIA Blackwell platform has arrived — enabling organizations everywhere to build and run real-time generative AI on trillion-param...

nvidianews.nvidia.com

5

1

24

Guido Appenzeller

@appenz

4 years

Today we are launching YubiEnterprise Delivery. You can now physically mail @Yubikeys to your employees or partners via a simple API call. Congrats to @Percy_Wadia , Mash and the team for getting this out the door.

Introducing YubiEnterprise Delivery

With YubiEnterprise Delivery, US and Canada-based organizations can ship YubiKeys directly to employees, partners, and contractors in more than 30 countries ...

www.youtube.com

0

12

24

Guido Appenzeller

@appenz

1 year

I now have an "Add Types" for Python. 🤯 🔥Hell just froze over 🧊 Generative AI is completely changing the developer experience. This is CoPilot (beta) + CoPilot Labs + VSCode Insider.

0

4

23

Guido Appenzeller

@appenz

1 year

Multi-modal models for OCR are ridiculously good.

1

24

Guido Appenzeller

@appenz

10 years

I just decided on my next adventure. And it’s not a startup. I am excited to join the best team in the entire networking industry!

14

21

23

Guido Appenzeller

@appenz

4 months

It was a privilege to hang out with some of the leading minds in AI art last night. Many artists appear to fear AI, but these individuals are not among them; they are boldly inventing the future of art.

Justine Moore

@venturetwins

4 months

Night 0 ( @a16z AI Artist Retreat)

10

5

110

2

6

23

Guido Appenzeller

@appenz

1 year

Dubbing videos in different languages is looking like a solved problem. The video below was recorded in 🇬🇧 English, and dubbed in 🇪🇸 Spanish. Below a description of the tech stack and the a list of models. Video from / @kn @fatimasabar 🧵Thread 1/6

2

4

23

Guido Appenzeller

@appenz

9 months

Apparently generating product names on Amazon with GPT-4 is a thing 🤣

3

15

23

Guido Appenzeller

@appenz

1 year

I fine-tuned SDXL on myself and I am really impressed. More detail, much more dramatic pictures, more consistency and good looking hands. Fine-tuning is still harder than SD 1.5. A few observations below. 1/5 🧵⬇️

4

23

Guido Appenzeller

@appenz

5 years

Great discussion on SIM Swaps. If there is one takeaway for me it is that no online service should use SMS as a 2nd factor.

Pemo Theodore

@pemo

5 years

Great enlightening discussion last nite Protect Yourself from Sim Swaps panel thx @appenz CPO @yubico @lrvick Security Researcher @michaelterpin CEO @TheBitAngels @boydrljr VP @THPII missed @Jon_A_Haas @BramahSystems thx @SheppardMullin

0

2

7

1

12

22

Guido Appenzeller

@appenz

7 years

(VM)ware -> (VM+Container)ware. #VMworld17 #VMWPKS

0

9

22

Guido Appenzeller

@appenz

8 years

VMworld Barcelona Keynotes now on Youtube

VMworld 2016 Europe - General Session Day 1

www.youtube.com

0

15

22

Guido Appenzeller

@appenz

2 years

The amount of hype around Generative AI is incredible, but is there real revenue? Turns out the answer is yes, and together with @BornsteinMatt and @casado we are trying to define the Generative AI stack and where value accrues.

martin_casado

@martin_casado

2 years

[New Post] Where does the value accrues in the generative AI stack? Most is accruing to infra, and there aren't clear signs of long term moats. With a market this size (all of human endeavor?) that's tremendously exciting. (w/ @BornsteinMatt , @appenz )

13

29

96

1

2

21

Guido Appenzeller

@appenz

1 year

3/7 First, the illustrative math mixes capex (GPU cost), annual opex, cumulative revenues during GPU lifetime & annual revenues from AI apps to derive the headline $200b number. A more appropriate calc would be based on the annual return earned on capital invested by GPU buyers.

2

1

22

Guido Appenzeller

@appenz

6 years

If in a platform centric future we do most networking at L7, we need the equivalent of TCP. This could be it.

Sébastien Deleuze

@sdeleuze

6 years

Other big announcement of the day at @s1p : #RSocket built by @facebook , @netifi_inc and @Pivotal . Reactive messaging with back pressure propagated to the clients over the network is a major game changer!

0

34

53

0

9

22

Guido Appenzeller

@appenz

5 years

If you plan to implement U2F on your site, take a look at this thread. Great list of anti-patterns to avoid.

Henri Sivonen

@hsivonen

5 years

I added U2F keys for a bunch of online accounts. So much badness in how these are implemented. Top annoyance: Requiring TOTP setup before you can add U2F keys. So if you thought that with a blue Yubico product (not full Yubikey), you'd avoid problems when switching phones, nope!

2

4

26

0

12

21