Explore tweets tagged as #DATALab
Everyone is sleeping on this new OCR model! Datalab's Chandra topped independent benchmarks and beat the previously best dots-ocr. - Support for 40+ languages - Handles text, tables, formulas seamlessly I tested on Ramanujan's handwritten letter from 1913. 100% open-source.
37
310
2K
نحوّل أي PDF أو صورة إلى مستند نصي عادي، لأن فريق Datalab نزل أفضل نموذج للتعرّف الضوئي على الحروف اسمه Chandra. ترمي الملف، ويعطيك النتيجة بصيغة HTML أو بصيغة Markdown أو بصيغة JSON. يطلع الجداول، والمعادلات، والرسومات البيانية بسهولة. يفهم أكثر من 40 لغة. في الاختبارات
10
105
861
🔥لازم تعرف عن الأداة اللي قلبت المو��زين! فريق Datalab أطلق نموذج OCR اسمه Chandra👀 يحوّل أي PDF أو صورة إلى نص عادي بدقة مذهلة، ويستخرج الجداول والمعادلات والرسومات البيانية بسهولة يدعم أكثر من 40 لغة تقدر تستخدمه مباشرة من المتصفح أو تثبّته محليًا والأجمل؟ مجاني تمامًا 🎯
5
149
1K
We shipped Chandra (our SOTA OCR model) but base latency wasn't good enough for production. So we trained an Eagle3 draft model: ✅3× lower p99 latency ✅40% higher throughput ✅zero accuracy loss Here's how we made Chandra OCR 3× faster with Eagle3 speculative decoding 🧵
3
6
36
ℹ️ For everyone’s information, here’s a sample case of confusion about how Naver DataLab works ⬅️ & the corrected version ➡️
1
102
328
📍 Simplified Naver tutorial so even the less informed can follow 💡 • Playground for you to test for yourselves: 🔗 https://t.co/13FVAD2UNp We understand this may be the very first time some have laid eyes on Naver DataLab - Naver’s official analytics site, thus we’re
ℹ️ For everyone’s information, here’s a sample case of confusion about how Naver DataLab works ⬅️ & the corrected version ➡️
1
91
233
Datalab just released their next-generation OCR model — Chandra!
2
0
7
1. 업비트 데이터랩(Upbit DataLab)에 알트코인 시즌 지수도 있네요. 5년 치 데이터입니다. 블록체인센터(Blockchaincenter)의 경우, 2017년부터 데이터가 있습니다. 각자 취향에 맞게 쓰면 됩니다. 알트코인 시즌 지수를 살펴보면, 평균 1년 마다 기회를 주긴했네요. 21년 이후 데이터를 살펴보면,
1
3
64
The Datalab API can now extract redlines and comments into clean markdown! This is great for analyzing legal documents with LLMs.
8
9
72
We’re teaming up with Operators & Friends, Build., and @pebble_bed to host Building for the Real World, a focused hack night in San Francisco. Every percentage point of productivity gain in manufacturing, logistics, and infrastructure compounds across the entire economy. AI has
0
4
11
I just completed the Data Engineer in Python track on @DataCamp and built my first ETL pipeline for a retail dataset alongside!🥳 You can check out the project using this link: https://t.co/iuER47CGke If you're also transitioning into DE, let's connectttt☺️
19
12
169
Extrêmement heureuse et fière de vous annoncer que le projet "Wo(rk) in Progress" qui vise à mettre en place une collection numérique des dessins de Wo - 500 à ce jour - a été retenu par la BNF dans le cadre des projets DataLab !
15
15
98
O DataLab Serasa Experian é patrocinador Silver do EmTech Brasil 2025! Referência em data science, o DataLab mostra como dados, tecnologia e inovação se conectam para impulsionar negócios com mais assertividade. Últimos ingressos em https://t.co/cqlKulwhgL
0
0
0
AI people, your dataset prep just got 10x faster! with Hyper-DataLab 📈 Turn any URL into training-ready datasets + interactive charts. Structured JSONL/CSV + clean visuals. Launching soon & powered by @hyperbrowser
1
3
13
We just shipped Track Changes Extraction for Word documents ✍️ You can now extract who changed what, when, and why directly into Markdown or HTML. So many legal teams and legal tech companies told us there wasn’t a reliable way to extract tracked changes at scale, so we're
1
1
8
做数据分析的人应该都懂那种崩溃感： Excel 一开就卡、SQL 跑半天、模型调参像瞎猜、报告一写就是一下午。 更别说很多人本职工作也不是“数据科学家”，但又不得不啃一堆 CSV、报表和图表。 这两天看到人民大学 RUC Datalab 开源的DeepAnalyze，有点眼前一亮。 它不是那种“帮你写点分析文案”的
5
35
176
Im #Vollbild-Beitrag des #SWR zu @TimKoffiziell & @Critical__Cat beurteilen keine Juristen geschweige denn Richter, ob Online-Kommentare Beleidigungen darstellen, sondern "Datenjournalistinnen" vom "SWR Datalab". "Was schon mal krass war[...] 145 Beleidigungen", verkündet Gina🧐
39
50
340
We just hit 93.9% on olmOCR—but we think the benchmark is saturated. olmOCR is still our favorite external benchmark and the @allen_ai team did fantastic work on it. But our latest models feel better in vibe checks yet score the same or worse. Turns out we're near the ceiling
2
2
35
Más de 5.000 expertxs impulsan biotecnología, diseño digital y sostenibilidad a través del DataLab, acelerando descubrimientos que mejoran la vida de las personas y del planeta. 🌍✨
0
0
0