Han Xiao @hxiao Twitter profile

Last Seen Profiles

@mklew629

@Jacksepticeye

@SafIsaev

@BanderoFashik

@ringo

@Iliketwice

@Argentina

@aledecobo

@InfernoRime

@Game42001

@CortSmith78

@_nickheath

@Klakala4

@ZachCalzada

@Allohioselect

@robinwriting

@TheMistressEnvy

@ziv_stahl

@ESCC_Chargers

@datechanspilowz

@HRHPBadSA

@littlelilith87

@jandakembangstw

@VWindependent

@kimjkxs

@nekomeow_uu

@annajay___

@therealkylesi11

@TiferetEvocatus

@8y0Hdf8f9cumX

@JustReginXD

@Majd_elarabi

@DonSpinell

@RioBaseball

@bader222abha

@srag82001764754

Han Xiao

@hxiao

2 years

Rationale是我司推出的一款专为管理者和决策者打造的分析成效工具，它集成了最新的GPT3.x和上下文学习(in-context learning)技术，能够快速生成Pros & Cons和SWOT分析报告，帮助管理者和个人做出明智的决策。 @mranti

24

130

427

Han Xiao

@hxiao

4 months

Here is an updated notebook for training Phi-3-tiny (66M parameters) from scratch on @MSFTResearch TinyStories dataset using @huggingface SFTTrainer. This is inspired by @astar_research STLM proposal. And it just works! With 24GB VRAM you can do batch size

Phi-3-TinyStories

Colab notebook

colab.research.google.com

4

79

363

Han Xiao

@hxiao

2 years

As everyone hails ChatGPT API, we had to speak up: our migration from davinci003 to gpt35-turbo actually made the generated content quality worse in many cases. While saving costs may be tempting, it's not worth sacrificing quality. Are we alone on this? #ChatGPT

40

32

309

Han Xiao

@hxiao

2 years

Watch me talk about the neural search ecosystem from @JinaAI_ . Since I founded the company in 2020, we have developed an #opensource product landscape for helping developers build deep learning-powered search applications. Follow me to know more!

The Neural Search Ecosystem from Jina AI

In this talk, Han gives an overview of the product landscape and the neural search ecosystem provided by Jina AI. If you have any questions, please follow us...

www.youtube.com

1

20

263

Han Xiao

@hxiao

6 months

Just dropped my DSPy intro slides from yesterday's presentation! They're designed with beginners in mind, featuring more visuals and less code to help you grasp the why and how of DSPy for prompt engineering. Also, I've cleared up some misunderstanding

6

58

269

Han Xiao

@hxiao

2 years

Excited to announce 🪩 𝗗𝗶𝘀𝗰𝗼𝗔𝗿𝘁: create compelling Disco Diffusion artworks in just one line! Radically easy, fully optimized for Google Colab free tier. @multimodalart #creativeai #generativeart #opensource

GitHub - jina-ai/discoart: 🪩 Create Disco Diffusion artworks in one line

🪩 Create Disco Diffusion artworks in one line. Contribute to jina-ai/discoart development by creating an account on GitHub.

github.com

1

20

214

Han Xiao

@hxiao

2 years

🤯 Build a "Zoom" in 20 lines of code with Jina and #Python . Check how: A great showcase of how powerful, easy, and efficient Jina is even for real-time streaming services. We will call it "Joom"! #opensource #neuralsearch #multimodal @tiangolo

1

28

241

Han Xiao

@hxiao

2 years

DALL·E Flow now has an official Docker image and requires only one GPU! Easy deployment! Thanks to the optimized diffusion step and @borisdayma new mega checkpoint, these are the fine artworks we get from DALL·E Flow! @multimodalart

4

29

214

Han Xiao

@hxiao

1 year

🎉 Unveiling 𝗝𝗶𝗻𝗮 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝘀! A new set of high-performance sentence embedding models, boasting between 35m to 6b parameters, expertly trained by @JinaAI_ and is accessible on 🤗. Excellent for neural search, reranking, and recsys 😎

Jina Embeddings: A Novel Set of High-Performance Sentence Embedding Models

Jina Embeddings constitutes a set of high-performance sentence embedding models adept at translating textual inputs into numerical representations, capturing the semantics of the text. These...

arxiv.org

4

64

202

Han Xiao

@hxiao

2 years

📣 Announcing 𝐏𝐫𝐨𝐦𝐩𝐭𝐏𝐞𝐫𝐟𝐞𝐜𝐭 - a prompt-engineer-centric tool that automatically optimizes your prompts for ChatGPT, GPT3.5+, DALLE, Stable Diffusion! Prompt engineering done right! #LLMs #LMOps #LMs

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

9

49

196

Han Xiao

@hxiao

4 months

@osanseviero 😬Poor Stanford undergrads, they would never thought they have inadvertently contributed to escalating tensions in the US-China AI race, geopolitical conflicts, and potential sanctions.

4

0

179

Han Xiao

@hxiao

2 years

Do you recognize this dataset? It is 𝗙𝗮𝘀𝗵𝗶𝗼𝗻-𝗠𝗡𝗜𝗦𝗧 created by me in 2017 when I was at @Zalando . Today, it just passed 10,000 @github stars and ~4,000 citations on @Google scholar!🎉 It has become one of the standard benchmark datasets in machine learning... 🧵

5

19

170

Han Xiao

@hxiao

4 months

hmm, this sounds too ambitious to me; but here is an interesting research proposal from @astar_research on Super Tiny Language Models (STLM) with only 10M, 50M, and 100M parameters. They aim to achieve competitive performance compared to models in the size range of 3B-7B

7

46

174

Han Xiao

@hxiao

4 months

Here is an untrained Phi-3-tiny with 50M parameters inspired by Super Tiny LM paper from @astar_research @LeonGuertler . In their original proposal, Llama2 was the base so the tokenizer & activation function r different. Feel free to experiment it

phi3-tiny-untrained.ipynb

Colab notebook

colab.research.google.com

3

28

164

Han Xiao

@hxiao

11 months

How did we beat OpenAI's text-embedding-ada002 on 8K token length? When and why 8K token length matters to embeddings? Read our paper released today

2

31

144

Han Xiao

@hxiao

2 years

Exciting news! PromptPerfect now offers auto prompt engineering for #GPT4 and @LexicaArt - I'm really impressed by GPT4 for its complex reasoning and math problem-solving! Let's see some examples 🧵

5

27

137

Han Xiao

@hxiao

6 months

OpenAI's ChatGPT exclusive "Breeze" voice is impressively lifelike, better than any voice option in their public API.

12

19

107

Han Xiao

@hxiao

5 months

People think LLMs hallucinate because of the knowledge cut-off, i.e., you are asking for new information created since training. This is true, but there is also another reason: you are asking for niche knowledge that has been "marginalized" during training. For example, when I

7

19

93

Han Xiao

@hxiao

2 years

Do you like our new @github organization page? 🤩Check it out: #opensource #neuralsearch

1

6

84

Han Xiao

@hxiao

6 months

I've been diving into DSPy last few days, and although it's impressively powerful, I've hit a few bumps trying to wrap my head around some concepts. Here's my take on it: 1⃣ What are teleprompter, optimization, and compile? What's exactly being optimized? In DSPy,

3

17

95

Han Xiao

@hxiao

3 years

What's a better way to spend Friday night to support FastAPI and pydantic? Many! But no regret.😂 DocArray 0.1.8 now supports data validation and now you can seamlessly use it in FastAPI to build reliable webservices! @tiangolo @samuel_colvin

4

10

86

Han Xiao

@hxiao

1 year

Sorry to be the buzz killer this #AutoGPT party. Here is my unpopular opinion about it. Today, I had a time to look at its source code and play it with my colleagues at @JinaAI_ , here is what I learned 👇

3

30

81

Han Xiao

@hxiao

6 months

@AustinTByrd need to delve into it.

2

1

80

Han Xiao

@hxiao

5 months

@samuel_colvin @NASA @pydantic @KPMG They kinda apply this guilty-until-proven-innocent logic to OSS, which is understandable in US but indeed very uncomfortable

3

0

80

Han Xiao

@hxiao

1 year

🧁PromptPerfect 0.10 has just been released! With Kandinsky 2.1 support, a new text2image model from @sberbank AI lab; and the new template management system! Try it now, and be impressed by Kandinsky's image quality and speed!

2

18

66

Han Xiao

@hxiao

2 years

While migrating from davinci003 to ChatGPT API @OpenAI released yesterday, we found two interesting observations. Good or bad? u tell me. First, the `assistant` role in the new API always addresses itself in the first person. This can be convenient in conversation UX, but in

3

11

61

Han Xiao

@hxiao

2 years

🙌Join 07/23 🇦🇹 Vienna. 𝐂𝐫𝐞𝐚𝐭𝐢𝐯𝐞 𝐀𝐈 & 𝐀𝐫𝐭 is the perfect event for AI engineers, artists and enthusiasts! This #Saturday , don't miss out on the chance to explore the intersection of creativity and artificial intelligence. #CreativeAI #Art

0

6

61

Han Xiao

@hxiao

2 years

Just attended @seb_ruder 's amazing tutorial on parameter-efficient fine-tuning at #EMNLP2022 and my mind is blown! His guidance has given me so many ideas for the next release of Finetuner. Thank you for sharing your wisdom, Seb!

1

4

57

Han Xiao

@hxiao

2 years

Friday night #DiscoArt with 🧵🎨 DocArray ID attached: 'discoart-1490247652'

5

8

52

Han Xiao

@hxiao

2 years

In their latest demo, @OpenAI unveiled the impressive multimodal capabilities of #GPT4 , generating text descriptions from images with ease. Give PromptPerfect 0.6 a spin to experience this feature firsthand! Spoiler: so much better than #BLIP2 ! Let's see some examples, 🚀🧵

4

18

54

Han Xiao

@hxiao

2 years

huh? ChatGPT Plus is intrinsic `text-davinci-002`??? Thought it should -004 or something, not even davinci-003? @OpenAI

6

4

53

Han Xiao

@hxiao

4 years

3 months ago, I announced my new venture on Neural Search. Today, I’m proudly revealing 🔍Jina: the cloud-native neural search framework powered by state-of-the-art AI & deep learning. Think out-of-the-[text]box, use Jina to build your next search system!

1

12

51

Han Xiao

@hxiao

7 months

@yetone “找不到对象”

2

1

50

Han Xiao

@hxiao

6 years

First post of 2019🎊: serving Google BERT model in production using Tensorflow and ZeroMQ, where I explain the design philosophy behind my open source project "bert-as-service". If your new year’s resolution is putting your ML/AI p…

This link will take you to a page that’s not on LinkedIn

lnkd.in

1

12

44

Han Xiao

@hxiao

2 years

LLM/ChatGPT hype today in one pic. If one really wants to build a decentralized LLM, 𝐭𝐡𝐞 𝐨𝐰𝐧𝐞𝐫𝐬𝐡𝐢𝐩 𝐨𝐟 𝐭𝐡𝐢𝐬 𝐋𝐋𝐌 𝐬𝐡𝐨𝐮𝐥𝐝 𝐛𝐞 𝐝𝐢𝐬𝐭𝐫𝐢𝐛𝐮𝐭𝐞𝐝 𝐭𝐨 𝐞𝐯𝐞𝐫𝐲 𝐩𝐞𝐫𝐬𝐨𝐧 𝐰𝐡𝐨 𝐜𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐞𝐬 𝐭𝐨 𝐭𝐡𝐞 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠 𝐜𝐨𝐫𝐩𝐮𝐬, even

5

11

44

Han Xiao

@hxiao

4 months

@BenjaminDEKR It’s true, but the idea of "training on YouTube" is a major misconception in the West. Here’s why: 1. Most Chinese users aren’t thrilled with Westernized GenAI image and video results because they lack cultural roots and elements. For example, they might wonder why a guy has

4

39

Han Xiao

@hxiao

2 years

So so much fun to practice Jina () in a GPU-heavy pipeline consisting of #DALLE -mega @borisdayma , GLID, SwinIR, and CLIP-as-service. Can't stop 😁 Early next week I will share a super-easy notebook for you to reproduce. 🧵

2

9

40

Han Xiao

@hxiao

1 year

🚀Unleash the power of visual storytelling with #SceneXplain 0.2! Introducing the new & fast "Comet" algorithm to solve hallucination issues & elevate your image narratives. Now with batch support for 128+ images in a single shot! Experience it today:

1

7

39

Han Xiao

@hxiao

1 year

We benchmarked @midjourney /describe command released earlier today vs. SceneXplain released yesterday vs. CLIPInterogator 2.1 and BLIP2 on image captioning and "reverse-engineering" prompts, here is what we learned. Full article

SceneXplain: Unleash the Advanced Image Captioning & Storytelling

Uncover the game-changing potential of SceneXplain, an advanced image captioning solution powered by LLMs. Check out the benchmark against Midjourney, CLIP, BLIP2, and other alternatives. Dive into...

jina.ai

2

20

38

Han Xiao

@hxiao

2 years

Super excited to announce DALL·E Flow: a Human-in-the-Loop workflow for creating HD images from text. Special thanks to @borisdayma @rom1504 for their support over the weekend! #opensource #dalle Now, look at the amazing results that you can get in 🧵

GitHub - jina-ai/dalle-flow: 🌊 A Human-in-the-Loop workflow for creating HD images from text

🌊 A Human-in-the-Loop workflow for creating HD images from text - jina-ai/dalle-flow

github.com

1

9

36

Han Xiao

@hxiao

1 year

@dotey 为什么不直接prompting给GPT？因为这里用到了Langchain实现的ReAct framework实现Agent之间推理和协同。我司上个月也有类似的项目

GitHub - jina-ai/agentchain: Chain together LLMs for reasoning & orchestrate multiple large models...

Chain together LLMs for reasoning & orchestrate multiple large models for accomplishing complex tasks - jina-ai/agentchain

github.com

4

15

34

Han Xiao

@hxiao

2 years

@fchollet Jina: because I’m the creator of it.

0

1

35

Han Xiao

@hxiao

4 months

@yangzhizheng1 @OpenBMB @zibuyu9 @yaoyuanAIR @imxiaohu @Datou @ZHOZHO672070 @Gradio @AkshGarg03 @siddrrsh i think they “apologized”?

3

35

Han Xiao

@hxiao

6 months

"Bootstrapped 0 full traces after 20 examples in round 0" is probably the most frustrating message for DSPy newbies. This silent error essentially means that optimization/compilation failed ❌, and the prompt you get is no better than simple few-shot. What goes wrong ? I've

7

8

33

Han Xiao

@hxiao

2 years

@oran_ge 这里的观点和我之前写的一篇文章一样，ChatGPT没有帮助SEO而是直接杀掉了SEO，当然取而代之的是LLMO，或者学术圈叫做in-context learning

SEO is Dead, Long Live LLMO

With ChatGPT replacing Google search, traditional SEO is dead: how can businesses ensure their information appears in ChatGPT's single answer? The solution is in-context learning on LLMs.

jina.ai

1

6

32

Han Xiao

@hxiao

4 months

It is hard to tell if people hate to love RAG or love to hate RAG. According to recent discussions on X and HN, RAG should be dead again. This time, critics are focusing on the over-engineering of most RAG frameworks, which, as @jeremyphoward @HamelHusain @Yampeleg demonstrated,

0

6

32

Han Xiao

@hxiao

2 years

@MicrosoftLoop clippy: that eyes are mine and give back my eyebrows!

1

32

Han Xiao

@hxiao

2 years

@oldj 直接用PromptPerfect 还可以对image2prompt逆向工程

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

11

31

Han Xiao

@hxiao

4 years

Sat.night 2 hours on porting @GoogleAI latest BiT model in (Poké-)production using @JinaAI_ . Now you can look for similar @Pokemon with SOTA visual representation learning. Feat. replicas and shards, containerization, REST & gRPC gateway, you name it!

1

9

32

Han Xiao

@hxiao

1 year

🚀Blast off with PromptPerfect 0.20! New onboarding is as easy as pie 🥧, more tutorial videos than you can shake a stick at! This isn't just an update; it's a whole level-up! Caution: side effects include exceptional PE skills🛠️ and irresistible charm. 😎

2

17

32

Han Xiao

@hxiao

1 year

🚀PromptPerfect 0.9 now supports auto prompt engineering for Claude from @AnthropicAI . Imo, Claude is a smart AI but often too "stuffy": too polite, serious, can't joke. Now PP loosened it up 😁Try it yourself.

1

16

31

Han Xiao

@hxiao

3 years

In Jina 2.0 (), we extend #FastAPI #SwaggerUI to pretty-print multimedia responses into listview & flowchart. This allows Jina developers to play and test REST endpoints in prompt. Try it now!

1

4

32

Han Xiao

@hxiao

2 years

SEO is dead as #ChatGPT is replacing Google search, so how can businesses ensure their info appears in its single answer? In-context learning on LLMs is the solution. Special thanks to @GaryMarcus @seb_ruder @Nils_Reimers for inspiration at @emnlpmeeting

SEO is Dead, Long Live LLMO

With ChatGPT replacing Google search, traditional SEO is dead: how can businesses ensure their information appears in ChatGPT's single answer? The solution is in-context learning on LLMs.

jina.ai

5

10

31

Han Xiao

@hxiao

2 years

Introducing 𝗥𝗮𝘁𝗶𝗼𝗻𝗮𝗹𝗲: a decision-making tool powered by GPT3.x and in-context learning for analyzing Pros & Cons, SWOT. Perfect for managers, business owners, and individuals. 🌍Multilingual support! 🌟Try it now #GPT3 #GenerativeAI

0

12

29

Han Xiao

@hxiao

6 months

My hot takes on DSPy, why it's crucial for future prompt engineering, and yet why it's challenging for average prompt engineers to learn.

Jina AI

@JinaAI_

6 months

Heads up, Bay Area guys ditched their AVP 👓 already and buzz about DSPy now. Could DSPy be the fresh go-to framework for prompt engineering after LangChain and LlamaIndex?

2

31

179

0

8

31

Han Xiao

@hxiao

1 year

📷Discover the power of #GPT4 -like multimodal image explaining with SceneXplain! Advanced image storytelling driven by LLMs, tailored for complex scenes & multilingual support. Fast batch processing with our API. Don't wait; elevate your visuals today!

SceneXplain - Leading AI Solution for Image Captions and Video Summaries

Experience cutting-edge computer vision with our premier image captioning and video summarization algorithms. Tailored for content creators, media professionals, SEO experts, and e-commerce enterpr...

scenex.jina.ai

2

19

30

Han Xiao

@hxiao

2 years

PromptPerfect 0.4 now improves the auto prompt engineering and supports @midjourney You can now easily generate stunning images 𝙬𝙞𝙩𝙝𝙤𝙪𝙩 carefully designed prompts.

1

9

29

Han Xiao

@hxiao

4 years

A big & solid 1.0 after a year of hard work! Congrats to our team and OSS community! ❤️ @PythonWeekly @ThePSF @PyData @pydataberlin @pycon @gvanrossum @pycoders @PythonHub @pypi @kdnuggets And a big thank you to our upstream: @numpy_team #Protobuf #gRPC @libzmq #PyYAML #FastAPI

Jina AI

@JinaAI_

4 years

🎉Today we are excited to announce Jina 1.0 — an easier way to build neural search on the cloud. 🌌 Universal search on image, text, audio ... ⚡ Lean & fast ⏱️ Time saver 🍱 Full-stack ownership 🧠 First-class AI models 🌩️ Cloud ready 👉 #opensource

4

55

188

0

11

29

Han Xiao

@hxiao

2 years

@gdb comparing to davinci003, ChatGPT API implementation of ethical filtering can be excessive, as it tends to include disclaimers and even censor friendly sarcasm. Should a competitor arise with a less restrictive alternative, this could be the downfall of ChatGPT API.

2

1

27

Han Xiao

@hxiao

3 years

Thanks to @willmcgugan #Rich , DocArray nesting and embedding visualization just got way cooler.

0

4

26

Han Xiao

@hxiao

6 months

Today, I visited a big tech in Bay, a key user of jina-embeddings-v2. Their feedback was encouraging: "We don't rely on public MTEB; we assess all embeddings on our domain data, and Jina is the best."🥹 At @JinaAI_ , we focus on solving real problems, not just meeting benchmarks.

2

27

Han Xiao

@hxiao

5 years

🎉Introducing my latest work: GNES!🍾GNES is Generic Neural Elastic Search, a cloud-native semantic search system based on DNN, it enables large-scale index and search for text-to-text🔠, image-to-image🖼️, video-to-video🎞️and any-to-any content form.

GitHub - gnes-ai/gnes: GNES is Generic Neural Elastic Search, a cloud-native semantic search system...

GNES is Generic Neural Elastic Search, a cloud-native semantic search system based on deep neural network. - GitHub - gnes-ai/gnes: GNES is Generic Neural Elastic Search, a cloud-native semantic s...

github.com

3

4

26

Han Xiao

@hxiao

2 years

Buying a Porsche 911🏎️ or a house in Spain🏡? Let 𝐑𝐚𝐭𝐢𝐨𝐧𝐚𝐥𝐞 0.2 choose the best for you. The new multi-option analysis uses in-context learning and latest #GPT3 .x to generate a multi-criteria analysis that ends your indecisiveness #GenerativeAI

0

27

Han Xiao

@hxiao

2 years

@KittenYang 只是模仿英文语音的话，推荐Coqui AI，柏林隔壁一家初创，创始人之前做Firefox TTS的。目前图形化界面和开源的都有。我司之前拿这个模仿过Morgan Freeman，效果你们说

Han Xiao

@hxiao

2 years

One of the most exciting features of Rationale: the "multiverse" mode. What if we can glimpse the 𝗺𝘂𝗹𝘁𝗶𝘃𝗲𝗿𝘀𝗲 of outcomes before making a decision? #LLM #ChatGPT #multiverse

3

2

20

2

5

26

Han Xiao

@hxiao

2 years

@CarsonYangk8s 也可以把复制出来的prompt放到作进一步的优化

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

5

24

Han Xiao

@hxiao

6 years

BERT-as-service is honored to be one of the most popular open source project of the past year. Thanks all for making it happen!🤜🏻🤛🏻🎊

Mybridge

@Mybridge

6 years

Amazing Machine Learning Open Source Tools & Projects of the Year (v.2019). #AI #DeepLearning

0

16

19

2

7

26

Han Xiao

@hxiao

2 years

Thanks @OpenAI . This time is quick ⏩⏩

3

1

26

Han Xiao

@hxiao

2 years

@xicilion 还是自然语言，直接走一层pp即可，不仅gpt, dalle, midjourney 都可以用

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

1

25

Han Xiao

@hxiao

1 year

🔥 SceneX vs. miniGPT4: New benchmark results reveal SceneX's superior performance in image captioning! 🚀 SceneX consistently outperforms miniGPT-4 in capturing intricate visual details and generating engaging captions. Full breakdown in this thread!👇

SceneXplain vs. MiniGPT4: A Comprehensive Benchmark of Top 5 Image Captioning Algorithms for...

Uncover the future of image captioning as SceneXplain and its rivals face off in an epic showdown. Explore their impact on accessibility, SEO, and storytelling, and dive into our intriguing results...

jina.ai

1

15

26

Han Xiao

@hxiao

7 months

If you still think cosine similarity between document embedding and query embedding for search, think again! Dive into ColBERT now:

Jina AI

@JinaAI_

7 months

Last Friday our 8192-length Jina-ColBERT on @huggingface and @bclavie has set Twitter abuzz. But why? And what is ColBERT anyway? This article unpacks ColBERT and ColBERTv2, explains their designs and why @lateinteraction is a game-changer for search.

1

8

50

1

3

25

Han Xiao

@hxiao

1 year

@abacaj Agree! using vector DB in this GenAI/Agent “looong memory” hype is mostly just overkill. Serly, how long can this memory be? Most seasoned wrapper apps have at most 4 digits “memory”, big deal, which can be easily done via exhaustive search in mem/browser.

1

2

23

Han Xiao

@hxiao

2 years

Back from @KubeCon_ and unpacking, I don’t think I need any socks 🧦 this summer. Thanks a lot for these creative and colorful swags 🌈 @DataStax @runailabs @elastic @InfluxDB @ocrasec @aiven_io @zesty_co @era #KubeCon #CloudNativeCon #KubeConEU

1

24

Han Xiao

@hxiao

2 years

@OfficialLoganK automatic prompt engineering

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

4

24

Han Xiao

@hxiao

1 year

@mranti Claude太stuffy了，用中文来讲就是有迂腐不够骚。直接向Claude提问往往会出发ethical filter被拒。不过可以通过PromptPerfect 优化Prompts后实现“越狱” 这里给出了一个让Claude写个Rap，它一开始还扭扭捏捏说不会写

Han Xiao

@hxiao

1 year

🚀PromptPerfect 0.9 now supports auto prompt engineering for Claude from @AnthropicAI . Imo, Claude is a smart AI but often too "stuffy": too polite, serious, can't joke. Now PP loosened it up 😁Try it yourself.

1

16

31

0

2

20

Han Xiao

@hxiao

1 year

🚀PromptPerfect 0.21 is out! Feel like a wizard as you brew complex few-shot prompts right from the cozy comfort of the UI. Fiddle with the instructions, upload examples, and fine-tune just the way you like it. Time to let your few-shot prompt genie out of the bottle!

4

5

22

Han Xiao

@hxiao

1 year

🐦 Introducing #JinaChat , the very own ChatGPT-like service from the creative minds @JinaAI_ ! We're talking multimodal chats🖼️, built-in memory recall🐘, and wallet-friendly rates💰. Its API is the perfect fit for building AutoGPT and agent-based apps.

JinaChat - More modality, longer memory, less cost

Multimodal chats, endless memory, and budget-friendly API to reshape how we communicate and create.

chat.jina.ai

1

12

21

Han Xiao

@hxiao

5 years

Check out my new blog post about building a Generic Neural Elastic Search system, moving from bert-as-service to a much more a generalized and elastic solution. @TDataScience @kdnuggets @TFBestPractices @icoxfog417

1

7

23

Han Xiao

@hxiao

1 year

Unleash the 🦍BEAST MODE🦍 in PromptPerfect 0.12! Put your optimized prompt to the test with our all-new Playground tab, featuring seamless compatibility with top-tier LLMs & text2image models. Give it a whirl & go viral! 🔥 Try it now: 🚀

1

6

21

Han Xiao

@hxiao

3 years

@JinaAI_ @tiangolo @willmcgugan And thanks to @willmcgugan for his smooth progress bar in rich library, making the user experience even more perfect!

0

1

22

Han Xiao

@hxiao

1 year

Had a fun weekend experimenting with @ciguleva 's Midjourney translucent prompt on some iconic cars: @Porsche 911, @Audi S8 , @MercedesBenz S-class & @BMW M4. 🔍🚗 Findings: - Making cars translucent is tough! The latest Porsche 911 (992) resisted my efforts, older models? A

2

6

21

Han Xiao

@hxiao

2 years

@ashu_mest 或直接用PromptPerfect生成，一边生成一边获得灵感和启发

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

1

4

20

Han Xiao

@hxiao

2 years

@midjourney PromptPerfect 0.4 released today now also supports auto prompt engineering for @midjourney Check it now

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

2

19

Han Xiao

@hxiao

6 months

@Korayem here u go

DSPy: Not Your Average Prompt Engineering

Heads up, Bay Area guys ditched their AVP already and buzz about DSPy now. Could DSPy be the new go-to framework for prompt engineering after LangChain and LlamaIndex?

jina.ai

2

1

21

Han Xiao

@hxiao

1 year

@omarsar0 ❤️Great collection! In the upcoming release of PromptPerfect, we're introducing a multi-agent playground. Dive into simulations where multiple agents interact, collaborate, or challenge each other to achieve their objectives. Here's a sneak peek.

0

4

21

Han Xiao

@hxiao

1 year

🚀Launching SceneXplain 0.4 with 2⃣ new image captioning algorithms: Ember🔥 & Flash⚡️! Ember slays those pesky summary sentences for crisp captions, while Flash zips through your scenes at lightning speed!💨 Try it now🌟

0

8

21

Han Xiao

@hxiao

3 years

@willmcgugan #Rich is awesome! Now using CLIP-as-service to embed images and texts into vectors via a GPU server. I always want to show the duplex streaming, QPS, data size, and progress in one place. Looks like I found a perfect solution!

0

1

21

Han Xiao

@hxiao

2 years

@gdb would be nice to have a token counting API.

4

0

21

Han Xiao

@hxiao

7 years

What a weekend! Fashion-MNIST dataset became the 3rd most-popular repo on GitHub Saturday. Just 24 hours after release!

0

1

16

Han Xiao

@hxiao

4 years

🐣 New to neural search? Curious about what makes Jina today? I made a curated list 📃 including the most featured soundbites from my earlier posts. Check it out!

From Then to Now: a Curated List for Neural Search and Jina · Han Xiao Tech Blog - Neural Search &...

Since 2017 I have been working on the neural search in different teams: Zalando Research (2014-2017), Tencent AI Lab (2018-2020), and now at Jina AI w ... · Han Xiao

hanxiao.io

1

4

19

Han Xiao

@hxiao

2 years

Excited to see my name on the giant screen! thanks @linuxfoundation #OSSummit

1

2

19

Han Xiao

@hxiao

2 years

One of the most exciting features of Rationale: the "multiverse" mode. What if we can glimpse the 𝗺𝘂𝗹𝘁𝗶𝘃𝗲𝗿𝘀𝗲 of outcomes before making a decision? #LLM #ChatGPT #multiverse

3

2

20

Han Xiao

@hxiao

2 years

@techreview Without providing an opt-out option, creative art is a copyright nightmare and an anarchist carnival. People just use “opensource” as a shield to dodge the data compliance issue.

0

1

20

Han Xiao

@hxiao

7 years

Still using #solr / #elasticsearch for product search? then don't miss my new blog post. I'll show you how to use #TensorFlow to build a cross-lingual end-to-end product search system.

Building Cross-Lingual End-to-End Product Search with Tensorflow · Han Xiao Tech Blog - Neural...

Product search is one of the key components in an online retail store. Essentially, you need a system that matches a text query with a set of products ... · Han Xiao

hanxiao.io

0

7

19

Han Xiao

@hxiao

1 year

🤓 𝗣𝗿𝗼𝗺𝗽𝘁 𝗔𝘂𝘁𝗼𝗖𝗼𝗺𝗽𝗹𝗲𝘁𝗲 is here! Since the first launch of PromptPerfect in March 2023, we've surpassed 𝟏𝟓𝟎,𝟎𝟎𝟎 𝐫𝐞𝐠𝐢𝐬𝐭𝐞𝐫𝐞𝐝 𝐮𝐬𝐞𝐫𝐬 and crafted over 𝟐.𝟐 𝐦𝐢𝐥𝐥𝐢𝐨𝐧 𝐩𝐫𝐨𝐦𝐩𝐭𝐬 in just a few months! PromptPerfect is quickly emerging as

1

7

19

Han Xiao

@hxiao

4 months

Okay so before Super Tiny LM, maybe we shall first look at Tiny LM from @MSFTResearch , published on May 24, 2023 (and @astar_research 's publishing date is exactly one year after MSFT, fun fact). In MSFT paper, they trained models with 1M, 2.5M, 8.3M, 28M, and 33M parameters

Han Xiao

@hxiao

4 months

hmm, this sounds too ambitious to me; but here is an interesting research proposal from @astar_research on Super Tiny Language Models (STLM) with only 10M, 50M, and 100M parameters. They aim to achieve competitive performance compared to models in the size range of 3B-7B

7

46

174

0

8

19

Han Xiao

@hxiao

1 year

How to describe spatial relations in one-dimension. Turn out there are 13 types of relations in 1D! 🤯 Source: J.F. Allen, Maintaining Knowledge about Temporal Intervals, Comm. ACM, vol. 26, pp. 832-843, 1983.

1

4

19

Han Xiao

@hxiao

1 year

@mranti 可以用中的“绕过道德过滤器”模式去改写原始prompt

PromptPerfect - AI Prompt Generator and Optimizer

Unlock prompt optimization for models like GPT-4, ChatGPT and Midjourney. Generate and refine prompts to perfection, receiving improved outcomes in seconds.

promptperfect.jina.ai

0

6

18

Han Xiao

@hxiao

2 years

Remember in #GPT4 release, @gdb demo "summarize everything using a sentence with words begin with G"? In PromptPerfect 0.7 you can add variables to a prompt and make it a prompt template! The optimized prompt will carry those variable placeholders 🤩

1

11

19

Han Xiao

@hxiao

4 months

Everyone hates hallucination. But Sycophancy is also a big issue here (the tendency of AI to agree with whatever you say), especially in a conversation. Sycophancy is a side effect of RLHF-like alignment training. I've noticed this behavior in many LLMs for a while, but only a

Stanford NLP Group

@stanfordnlp

4 months

Do the RAG GenAI systems of the big law tech companies give the “hallucination-free” drafting assistance they claim? @chrmanning & Mirac Suzgun aided a @Stanford RegLab study showing they do much better than GPT-4 but still hallucinate about 1 time in 6.

4

27

98

3

0

19

Han Xiao

@hxiao

2 years

@oran_ge 算法功课做的差有很多种，但百度这种功课做的考官看了都能笑出声来，也算负负得正吧。

1

0

14