正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Can Google Gemma DiffusionGemma help fix broken OCR? In theory, denoising tokens in parallel could work better for OCR correction since context is seen upfront? Pointed it at 19th-century newspaper OCR. It corrected better than the autoregressive baseline — at ~8x the speed.

Daniel van Strien

5,732 subscribers

36,105 次观看 • 1 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Unlimited-OCR 🔥New OCR from PaddlePaddle It can parse hundreds of pages in a single pass while maintaining stable speed. The key idea is R-SWA (Reference Sliding Window Attention), which keeps KV cache constant during decoding. 🏆 93% on OmniDocBench 📈 +6% over DeepSeek-OCR

Unlimited-OCR 🔥New OCR from PaddlePaddle It can parse hundreds of pages in a single pass while maintaining stable speed. The key idea is R-SWA (Reference Sliding Window Attention), which keeps KV cache constant during decoding. 🏆 93% on OmniDocBench 📈 +6% over DeepSeek-OCR

Adina Yakup

288,872 次观看 • 1 个月前

I just built my own OCR app powered by Google's Gemma 3. It runs 100% locally on my computer using Ollama and extracts text from images as structured markdown. The code is open-source, and you can find it in the next tweet.

I just built my own OCR app powered by Google's Gemma 3. It runs 100% locally on my computer using Ollama and extracts text from images as structured markdown. The code is open-source, and you can find it in the next tweet.

Akshay 🚀

477,685 次观看 • 1 年前

Today's autoregressive models generate one token at a time. Mercury 2 generates tokens in parallel. Over 1,000 tok/sec on standard GPUs, at comparable quality to speed-optimized models. Since launch, the community has been showing what diffusion LLMs can unlock. Thanks to the team at Clyep for the breakdown.

Today's autoregressive models generate one token at a time. Mercury 2 generates tokens in parallel. Over 1,000 tok/sec on standard GPUs, at comparable quality to speed-optimized models. Since launch, the community has been showing what diffusion LLMs can unlock. Thanks to the team at Clyep for the breakdown.

Inception

21,231 次观看 • 2 个月前

Mistral OCR 3 now returns confidence scores, at page level or word level. Know exactly how sure the model is about what it extracted. Available now via API and to test in Mistral Studio.

Mistral OCR 3 now returns confidence scores, at page level or word level. Know exactly how sure the model is about what it extracted. Available now via API and to test in Mistral Studio.

Mistral AI for Developers

29,944 次观看 • 3 个月前

I'm about to piss off a 13 billion dollar industry - the OCR market. Here's why: Monday: I optimized my PDF-to-Excel converter landing page for SEO. Tuesday: I started ranking #1 organically for "pdf to excel ai", a phrase with 8K+ monthly searches and growing. The Google Ads competition is insane. Wednesday: I was previously using a basic OCR API and I replaced it with the Gemini 2.0 Flash model, which is insanely good at PDF parsing. Thursday: I bought the domain pdftoexcelai[dot]com.

I'm about to piss off a 13 billion dollar industry - the OCR market. Here's why: Monday: I optimized my PDF-to-Excel converter landing page for SEO. Tuesday: I started ranking #1 organically for "pdf to excel ai", a phrase with 8K+ monthly searches and growing. The Google Ads competition is insane. Wednesday: I was previously using a basic OCR API and I replaced it with the Gemini 2.0 Flash model, which is insanely good at PDF parsing. Thursday: I bought the domain pdftoexcelai[dot]com.

David Bressler 📊

12,418 次观看 • 1 年前

People are overlooking Google Gemini Realtime models for computer use It gave me sub 100ms latency with computer use. It has a much larger context window and is much cheaper as well Combine that with local OCR and local screen detection model based on Omniparser by Microsoft it works under 100ms action taking when combined with Cua I also put in a harness for Nous Research Hermes with it. You can access it all at your tip of your cursor. You can draw on your screen to give your agents a context And I am making it Open Source! Link in the Comments Sundar Pichai Min-Liang Tan Mojtaba Seyedhosseini

People are overlooking Google Gemini Realtime models for computer use It gave me sub 100ms latency with computer use. It has a much larger context window and is much cheaper as well Combine that with local OCR and local screen detection model based on Omniparser by Microsoft it works under 100ms action taking when combined with Cua I also put in a harness for Nous Research Hermes with it. You can access it all at your tip of your cursor. You can draw on your screen to give your agents a context And I am making it Open Source! Link in the Comments Sundar Pichai Min-Liang Tan Mojtaba Seyedhosseini

Milind S

38,129 次观看 • 2 个月前

Fast tokens are the most valuable tokens. Because inference speed is productivity. And Cerebras makes the fastest tokens in the world. Google DeepMind's Gemma 4 runs at 1,500+ tokens per second on Cerebras — more than an order of magnitude faster than on GPUs and 15x faster than Claude Haiku 4.5. Private preview today. GA later this month. Build blazing fast multimodal apps.

Fast tokens are the most valuable tokens. Because inference speed is productivity. And Cerebras makes the fastest tokens in the world. Google DeepMind's Gemma 4 runs at 1,500+ tokens per second on Cerebras — more than an order of magnitude faster than on GPUs and 15x faster than Claude Haiku 4.5. Private preview today. GA later this month. Build blazing fast multimodal apps.

Andrew Feldman

12,008 次观看 • 1 个月前

Jordan Chiles deserved better than this. The point of any review in sports is to get it right. The judges didn't at first, then they corrected it. Now, because of 4 seconds past an arbitrary deadline that is different than the time her opponents had, Chiles medal is in jeopardy

Jordan Chiles deserved better than this. The point of any review in sports is to get it right. The judges didn't at first, then they corrected it. Now, because of 4 seconds past an arbitrary deadline that is different than the time her opponents had, Chiles medal is in jeopardy

Mike Leslie

563,309 次观看 • 2 年前

A vibrant Ghanaian youth has urged the youth to embrace a 'stay and fix' mindset and to work together in advocating for a better country, rather than abandoning it in search of better opportunities abroad.

A vibrant Ghanaian youth has urged the youth to embrace a 'stay and fix' mindset and to work together in advocating for a better country, rather than abandoning it in search of better opportunities abroad.

SIKAOFFICIAL🦍

126,484 次观看 • 2 年前

BRAD GERSTNER: “There's nobody better on planet Earth than Elon at converting electrons to tokens … He's monetizing it in a big way. It's terrific for Anthropic and it solves what I think was the biggest question in the valuation story.”

BRAD GERSTNER: “There's nobody better on planet Earth than Elon at converting electrons to tokens … He's monetizing it in a big way. It's terrific for Anthropic and it solves what I think was the biggest question in the valuation story.”

Chief Nerd

22,015 次观看 • 2 个月前

DrLupo says Marathon is good. He’s known about the project since 2018 and has been playtesting it for the past 6 years. Every time he’s seen it, it’s gotten better, and he’s very excited for people to finally see it. 👀 "I think that it can be better than what it is right now."

DrLupo says Marathon is good. He’s known about the project since 2018 and has been playtesting it for the past 6 years. Every time he’s seen it, it’s gotten better, and he’s very excited for people to finally see it. 👀 "I think that it can be better than what it is right now."

Marathon Bulletin

526,583 次观看 • 1 年前

Wow Google really cooked with the Gemini live API! Audio/Video reasoning at the speed of light ⚡️ The second answer is actually off but I'm amazed at how quickly it can "think". It will only get better from here. Gradio demo on Hugging Face 🔜

Wow Google really cooked with the Gemini live API! Audio/Video reasoning at the speed of light ⚡️ The second answer is actually off but I'm amazed at how quickly it can "think". It will only get better from here. Gradio demo on Hugging Face 🔜

Freddy A Boulton

19,971 次观看 • 1 年前

Enten: "This is the best looking polling for Texas Democrats in a Senate race since really the 20th century ... Talarico is doing 9 points better than O'Rourke [in 2018] ... this could be very, very real. Democrats have a real shot in Texas."

Enten: "This is the best looking polling for Texas Democrats in a Senate race since really the 20th century ... Talarico is doing 9 points better than O'Rourke [in 2018] ... this could be very, very real. Democrats have a real shot in Texas."

Aaron Rupar

77,933 次观看 • 3 天前

Running DiffusionGemma locally (26B-A4B, Q8_0 quant) was easy to set up. If you want to try it, you can follow the steps in the blog below. More testing will continue tomorrow. Note: The video is playing at 2x speed.

Running DiffusionGemma locally (26B-A4B, Q8_0 quant) was easy to set up. If you want to try it, you can follow the steps in the blog below. More testing will continue tomorrow. Note: The video is playing at 2x speed.

AshutoshShrivastava

11,169 次观看 • 1 个月前

holy shit, i built a better Cluely. introducing: Obscura, a better, more performant Cluely. Obscura helps you during your meeting and interview calls, in real time, with more context and a built-in agent to websearch in less than a second. it is completely INVISIBLE to your screen recorder, screen share and cannot be detected during your google calls or zoom calls no matter what. some functionalities: 1. toggle visibility: so that only you can see it 2. panic kill: if you wanna kill it in a second 3. screenshot ocr: for screenshot capabilities 4. context window: pass any context in .md or .docx file for it to read before your calls and Obscura will understand it 5. performative built-in agent: to search for anything in less than a second. it uses Deepgram for STT model and Anthropic / OpenAI api keys to generate answers. it takes $1 for 20 hours of your calls as compared to cluely's $20/mo or $75/mo plan (but that depends on tokens consumed on both models). releasing a .dmg for people for people to use. access will be given to people who share this with their friends. I will roll out a referral program so that people can share it with others. feel free to give some feedback. thanks to rahat for being on call with me and testing it.

holy shit, i built a better Cluely. introducing: Obscura, a better, more performant Cluely. Obscura helps you during your meeting and interview calls, in real time, with more context and a built-in agent to websearch in less than a second. it is completely INVISIBLE to your screen recorder, screen share and cannot be detected during your google calls or zoom calls no matter what. some functionalities: 1. toggle visibility: so that only you can see it 2. panic kill: if you wanna kill it in a second 3. screenshot ocr: for screenshot capabilities 4. context window: pass any context in .md or .docx file for it to read before your calls and Obscura will understand it 5. performative built-in agent: to search for anything in less than a second. it uses Deepgram for STT model and Anthropic / OpenAI api keys to generate answers. it takes $1 for 20 hours of your calls as compared to cluely's $20/mo or $75/mo plan (but that depends on tokens consumed on both models). releasing a .dmg for people for people to use. access will be given to people who share this with their friends. I will roll out a referral program so that people can share it with others. feel free to give some feedback. thanks to rahat for being on call with me and testing it.

Naruto11.eth

15,760 次观看 • 4 个月前

🚨 UNCANNY VALLEY: “HOPE YOU’RE BETTER AT IT THAN AI” Ethan Mollick has a message for every parent—and a warning about the future of work. Professor at The Wharton School, Ethan Mollick: “Your best bet is to do what you love—and hope that you’re better at it than AI. The question is, if AI is better than us at every skill… what do we do all day?” Uncanny Valley with Danish — only on 𝕏.

🚨 UNCANNY VALLEY: “HOPE YOU’RE BETTER AT IT THAN AI” Ethan Mollick has a message for every parent—and a warning about the future of work. Professor at The Wharton School, Ethan Mollick: “Your best bet is to do what you love—and hope that you’re better at it than AI. The question is, if AI is better than us at every skill… what do we do all day?” Uncanny Valley with Danish — only on 𝕏.

Mario Nawfal

31,347 次观看 • 1 年前

OpenRead is now open for PDF uploading and parsing, check it out! PDF upload parsing is one of the most desired features of our users, and now we can finally bring this futuristic technology to life. We have adopted the most advanced #NLP, #CV, #OCR technology, which can extract the text, formula, table, and figure in PDF to the greatest extent, and re-typeset these elements.

OpenRead is now open for PDF uploading and parsing, check it out! PDF upload parsing is one of the most desired features of our users, and now we can finally bring this futuristic technology to life. We have adopted the most advanced #NLP, #CV, #OCR technology, which can extract the text, formula, table, and figure in PDF to the greatest extent, and re-typeset these elements.

OpenRead

26,656 次观看 • 3 年前

Our default document parsing mode is now able to parse a complex research report with multiple embedded charts on a single page. This is the cheapest document OCR model out there that can turn complex visual documents into LLM-ready markdown. This is our agentic mode in LlamaParse. It starts at ~1c per page; if you’re looking to scale consumption we offer volume discounts. Come check it out! Sign up:

Our default document parsing mode is now able to parse a complex research report with multiple embedded charts on a single page. This is the cheapest document OCR model out there that can turn complex visual documents into LLM-ready markdown. This is our agentic mode in LlamaParse. It starts at ~1c per page; if you’re looking to scale consumption we offer volume discounts. Come check it out! Sign up:

Jerry Liu

15,479 次观看 • 6 个月前

The ‘E’ in DOGE is efficiency. It's not elimination. It's not extinction. We are trying to make the government do better for the American people and right-size it— the government can do a better job for the American people at a better price.

The ‘E’ in DOGE is efficiency. It's not elimination. It's not extinction. We are trying to make the government do better for the American people and right-size it— the government can do a better job for the American people at a better price.

Treasury Secretary Scott Bessent

22,395,999 次观看 • 1 年前

You could travel the world looking, but you won’t find many better par 3s than the sixth at Cruit Island in Donegal. It was looking spectacular today in the evening sun.

You could travel the world looking, but you won’t find many better par 3s than the sixth at Cruit Island in Donegal. It was looking spectacular today in the evening sun.

UK Golf Guy

35,273 次观看 • 3 年前