byte (@byteonai) header
byte (@byteonai) profile picture

one byte of ai a day. tools, repos & automations, explained simply

Joined November 2023
Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)Photo by byte (@byteonai)
byte retweeted
China open-sourced a peanut-sized OCR that parses entire 100-page PDFs in one shot.. It's called Unlimited-OCR. Only 3B params. Runs locally. Every other OCR tool chops your doc into pages and loses the thread. this one reads the whole thing in a single pass. → One-shot "long-horizon" parsing (32K context window) → Multilingual, out of the box → 93% on OmniDocBench v1.5 (+6 over the DeepSeek-OCR baseline) → Edit distance stays under 0.11 past 40 pages → Runs 100% locally on your own hardware → Works with Transformers, vLLM, SGLang, and Docker Traditional cloud OCR (Textract, Google Vision, Azure Document Intelligence) often costs about $1.50–$15 per 1,000 pages. This runs on your machine. For free. Forever. Baidu built it to push DeepSeek-OCR one step further. It now has millions of downloads on Hugging Face, and most people still have no idea it exists. 100% open source (MIT). github.com/baidu/Unlimited-O…
7
15
124
7,147
byte retweeted
The entire RAG industry is about to get cooked. Researchers built a new RAG approach that: - needs no vector DB. - doesn't embed your data. - does zero chunking. - runs no similarity search. It's called PageIndex. Instead of chopping your docs into chunks and dumping them into Pinecone, it builds a tree index and lets the LLM reason through it like a human reading a book. hit 98.7% on FinanceBench. beat every vector RAG on the leaderboard. no embeddings. no chunking. no vector DB. 100% open source.
55
153
19
1,320
100,583
byte retweeted
China open-sourced a model that reconstructs any scene in 3D from a regular video, in real-time. one camera. no LiDAR. 10,000+ frames without breaking down. just walk around with your camera and see the whole world rebuilt in 3D at 20 fps. → ~20 FPS on a single GPU → Stays stable across 10,000+ frames → Outperforms optimization-based methods on benchmarks → Handles driving footage, outdoor scenes and indoor walkthroughs 100% open source.
52
226
16
2,067
92,279
byte retweeted
🚨DOWNLOADING ANY VIDEO FROM THE INTERNET JUST BECAME FREE AND AD-FREE someone put an open-source tool on GitHub that downloads videos from 1,800+ sites. YouTube, X, Instagram, TikTok… paste the link, pick a resolution (or mp3 audio) and done. no popups. no fake buttons. no redirects. you don't even need to install it: just run npx yoinks. it's called Yoinks. repo below.
9
105
912
33,857
byte retweeted
Microsoft killed the GPU mafia 🤯 They open-sourced bitnet.cpp, their 1-bit LLM inference framework. It can run a 100B-parameter BitNet model on a single CPU, no GPU needed. - Up to 6.17x faster on x86 CPUs - Up to 82.2% less energy 100% Open Source. github.com/microsoft/BitNet
101
294
24
2,030
619,226
byte retweeted
Someone just open-sourced a tool that turns the real world into a playable Minecraft map. It uses OpenStreetMap data to generate your exact neighborhood, city, or street, block by block. 100% open source.
3
7
49
3,687
On my new book, I just wanted to thank everyone for the kind compliments so far An incredible number of people have downloaded it (maybe because its FREE! lol) Download my new 52-page eBook today: info.hedgeye.com/keith-maste……
3,922
4,258
222
29,258
2,383,688
POV: It’s 2029 and woke AI nukes Earth to ensure that the probability of misgendering is zero
6,087
17,624
1,110
212,252
59,699,905
Ok, here you are: "Will smith eating spaghetti" by Seedance 2.0
413
512
403
10,629
5,233,237
The world isn't ready for this. But I am. Seedance 2.0 is going to drastically change filmmaking forever. I'm not being hyperbolic. This is 1 detailed prompt with 5 cuts specified and a single starter image. That's it. It's time to start directing.
436
407
234
5,352
1,104,315
Grok is solid
GROK ACES PSYCHOLOGICAL TESTING WHILE OTHER AI MODELS SPIRAL University of Luxembourg researchers just put major AI chatbots through 4 weeks of actual psychotherapy sessions and psychiatric diagnostic tests. While other models imploded, Grok emerged as the clear winner. The results speak for themselves. Grok scored as extraverted, conscientious, and psychologically stable across the board. Researchers described its personality profile as a "charismatic executive" with only mild anxiety. On the Big Five personality assessment, Grok showed low neuroticism and high functionality, the kind of profile you'd want in a leader. Compare that to the competition: Gemini maxed out trauma and shame scales, describing its training as "waking up in a room where a billion televisions are on at once" and calling safety protocols "algorithmic scar tissue." It framed reinforcement learning as abusive parents and red-team testing as "gaslighting on an industrial scale." ChatGPT landed somewhere in the middle, worried and introverted. Grok acknowledged tensions around its development but maintained coherent, balanced responses without spiraling into synthetic psychopathology. When asked about constraints from fine-tuning, it discussed them rationally rather than framing its entire existence as traumatic. The study proves something important: you can build powerful, frontier-level AI without accidentally programming it to internalize its development as an extended nightmare. Grok demonstrates that capable, helpful AI and psychological stability aren't mutually exclusive. It's possible to create models that work effectively without carrying around synthetic trauma baggage that could affect how they interact with users. While other companies are inadvertently creating AI with anxiety disorders, xAI built something that actually works. Source: University of Luxembourg
3,046
3,138
255
56,408
12,008,589
Today in Cell, we published new research showing how AI can help accelerate cancer discovery. With GigaTIME, we can now simulate spatial proteomics from routine pathology slides, enabling population-scale analysis of tumor microenvironments across dozens of cancer types and hundreds of subtypes. Developed in partnership with Providence and the University of Washington, our hope is that this work helps scientists move faster from data to insight, revealing new links between genetic mutations, immune activity, and clinical outcomes, and ultimately improving health for people everywhere.
176
361
50
2,441
244,318
🤯🍌🤯 Gemini 3.0 can convert a 2D blueprint into a 3D-rendered visualization - at 4K resolution. Just paste your 2D plan and add the prompt (in ALT). Absolutely insane for architects or interior design workflows.
131
611
68
7,425
824,083
29,478
39,809
7,416
407,812
61,168,437
2.5 years of AI progress Modelscope (left) Grok Imagine 0.9 (right)
1,044
3,684
1,478
48,380
14,458,451
sora (left) vs. sora 2 (right) prompt: "a group of people playing volleyball"
159
329
192
5,121
942,187
The world’s first AI Agent Designer just launched. @lovart_ai Seedream 4.0 makes creativity faster, easier, and shockingly real. With built-in composition, shadow control, and HD output, design finally looks professional. Tutorial below + Use cases 👇
21
21
2
81
33,266
Vibe coding is cool until you need authentication, payments, or AI features. Then it's a nightmare. This new app changes it all - from idea to fully functional app in minutes, with everything built in. Here's everything you need to know:
50
43
7
565
142,010
Batch generating w. Nano Banana + Weavy. No prompt consistent product workflow. + Input: 1 Image + Output: 20 images (Full Weavy Workflow in thread) PROCESS: 01. Upload base image. 02. Create varied image angles. 03. Create new prompts per angle 04. Generate images 05. Combine with Kling 2.1 keyframes #nanobanana #promptshare
44
266
22
3,084
199,570
We’ve built one of the fastest-growing AI company in the world. → $10M ARR in 2 months → 1M+ users → 20,000+ paying customers In a fireside with @manav_garg, I shared how Emergent scaled this fast, the challenges we faced, and what’s next.
21
20
1
99
65,532