Xiaoyi Zhang
@xiaoyiz_uw
Followers
2K
Following
299
Statuses
62
HCI+AI @AnthropicAI | ex-Apple Staff Researcher, enabled on-device ML for Vision Pro & Intelligence | @uwcse PhD | Accessibility | Culinary Arts
Seattle
Joined July 2010
I'm super excited to join @AnthropicAI (Seattle office) for HCI + AI research with product impact! Look forward to contributing to responsive & safe AI! At @Apple, I did my first SWE internship (freshman) and 6 years later, joined as a researcher after completing my PhD. Grateful for the opportunity to lead 3 projects: 𝗦𝗰𝗿𝗲𝗲𝗻 𝗥𝗲𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻 empowers 20M+ users with visual impairments; 𝗧𝗮𝗹𝗮𝗿𝗶𝗮 ensures all Vision Pro ML models run efficiently; and all 𝗔𝗽𝗽𝗹𝗲 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 models are auto-compressed via our pipeline for amazing user experience on iPhone and server-side.
29
22
1K
Race to the top! Super excited about this launch and kudos to everyone working on this
Introducing the Model Context Protocol (MCP) An open standard we've been working on at Anthropic that solves a core challenge with LLM apps - connecting them to your data. No more building custom integrations for every data source. MCP provides one protocol to connect them all:
0
0
14
🎉
Claude 3.5 Haiku is now available on our API, Amazon Bedrock, and Google Cloud's Vertex AI. Haiku is fast and particularly strong at coding. It outperforms state-of-the-art models—including GPT-4o—on SWE-bench Verified, which measures how models solve real software issues.
2
0
11
0
0
1
👏
Claude is the best LLM in the world right now and the gap is not even close anymore. On Perplexity, you get to pick whatever model you want, be it Claude or Grok, or Llama 3.2 post-train or 4o.
0
0
5
RT @deedydas: Claude's Analysis mode > ChatGPT Code Interpreter. Why? Not restricted to python matplotlib / seaborn charts, but can genera…
0
74
0
Wow
Introducing, Act-One. A new way to generate expressive character performances inside Gen-3 Alpha using a single driving video and character image. No motion capture or rigging required. Learn more about Act-One below. (1/7)
0
0
1
Computer Use!
Introducing an upgraded Claude 3.5 Sonnet, and a new model, Claude 3.5 Haiku. We’re also introducing a new capability in beta: computer use. Developers can now direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking, and typing text.
0
0
14
RT @AnthropicAI: We've published a significant update to our Responsible Scaling Policy, which matches safety and security measures to an A…
0
136
0
@aaron_begg @AnthropicAI @Apple Thanks Aaron! Really glad that I had a chance to talk with you, which really helped me make a decision to join! Looking forward to working together (and see u soon in SF)!
0
0
6
Wanna thank @jeffbigham and @foil who brought me to Apple, the best place to work on accessibility😉 And thank Qi on AVP and AI product impact! Working with great talents, we published 2 best papers and a best paper nominee🏆 What a wonderful journey!
0
0
30
RT @BradSmi: The next episode of #ThePrompt features Saqib Shaikh and the Seeing AI app. This app uses a smartphone camera and generative A…
0
20
0
@_jgoertler @fredhohman @domoritz @donghaoren The Github repo: Really enjoy the amazing collaboration with Apple colleague @fredhohman @domoritz @donghaoren! Easter egg - import * as myc from '@apple/mycelium'; myc is the abbreviation of my wife's name 😉
1
1
8
RT @AnthropicAI: Today, we're making Artifacts available for all Claude users. You can now also create and view Artifacts on the Claude iOS…
0
333
0
Longer version tech report with more details 👍
As Apple Intelligence is rolling out to our beta users today, we are proud to present a technical report on our Foundation Language Models that power these features on devices and cloud: 🧵
1
0
9
Excited to share that our Talaria paper won Best Paper Nominee at #CHI2024! It is used in #WWDC24 #AppleIntelligence model optimization. Great teamwork with @fredhohman, C Wang, J Lee, @_jgoertler @domoritz, @jeffbigham, Z Ren, C Foret, Q Shan
1
7
38