Deep Learning Engineer . I train VLMs for document understanding
Bengaluru, India
- Tweets 64
- Following 87
- Followers 54
- Likes 109
Opus 4.7 토큰 테스트
토크나이저 차이로 제미나이의 2배를 사용합니다.
Opus 4.6 대비해서도 50% 많이 사용해요.
이건 사실상 같은 한도에서 모델이 50% 더 비싸진겁니다.
Community note
Ratio = 5,657 / 4,262 ≈ 1.327 → ~33% more tokens (not 50%).
x.com/midego1/status…
85
227
2,560
251,483
Maybe Claude code open sourced its codebase because it cannot figure out all the bugs related to high token usage and open sourcing it is more efficient. 🚀
1
92
Souvik Mandal retweeted
"When a metric becomes the target, it stops being a good metric" - Goodharts Law
last few days GLM-OCR has been trending after it claimed 95% on OmniDocBench, which is higher than Gemini-3-pro
in reality GLM-OCR is way worse than the story these benchmarks paint, lets see how
full disclaimer: ive been working in this space for the last 7 years with @nanonets
🚨BREAKING: A dev just open-sourced the #1 ranked OCR model on Earth.
It's called GLM-OCR and it just hit 94.62 on OmniDocBench V1.5, beating every OCR model in existence.
Only 0.9B parameters. One pip install. Handles documents no other model could touch.
100% Open Source.
1
10
12
543
Souvik Mandal retweeted
Every student accepted into Startup School India now gets $25k+ in AI and cloud credits.
Apply, get in, and start building: events.ycombinator.com/yc-su…
140
146
1,623
390,925
Souvik Mandal retweeted
A tiny phishing trick broke 5 out of 8 popular OCR models (Deepseek-OCR, Gemini-2.5Pro, PaddleOCR, etc)
Can you spot the phish in the screenshot?
If you use OCR/Markdown pipelines in prod, check out the results in the thread.
1/n
open-source OCR models are super cheap to run and privacy first 🤝
BUT there's a ton of new models out there: DeepSeek-OCR, Nanonets, PaddleOCR, how do you pick them? 🤯
don't worry though, @huggingface got you covered! 🫡🧶
12
21
147
42,734
Souvik Mandal retweeted
> RTX Pro 6000 96GB @PrimeIntellect
> Nanonets-OCR @nanonets
> vLLM @vllm_project
> 2000 images from cifar10 @huggingface
one word, damn!
18
23
338
230,149
Souvik Mandal retweeted
Nanonets released a new version of their SoTA OCR model 🔥
Supports LaTeX, Multilingual, Complex tables and much more
Works out of the box with transformers, vllm and all major runners! 🤗
huggingface.co/nanonets/Nano…
4
62
472
29,833
Live demo: docstrange.nanonets.com/
HF: huggingface.co/nanonets/Nano…
Blog: nanonets.com/research/nanone…
1
150
Souvik Mandal retweeted
After 1M+ downloads of their previous model, @nanonets is launching Nanonets-OCR2, a next-gen suite for image-to-markdown conversion & visual QA.
From LaTeX math & tables → mermaid charts, signatures → watermarks. It handles it all.
huggingface.co/nanonets/Nano…
4
8
93
22,402
We're working on hosting the Nanonets-OCR-s — you'll soon be able to run it directly using DocStrange with just a free API key!
Your LLMs are hungry for data, but documents are messy 😩.
DocStrange is the answer!
Our open-source solution turns any document into clean, LLM-ready data with one command. Give your models what they need.
🔗 github.com/NanoNets/docstran…
🔗 pypi.org/project/docstrange/
2
87
Souvik Mandal retweeted
Your LLMs are hungry for data, but documents are messy 😩.
DocStrange is the answer!
Our open-source solution turns any document into clean, LLM-ready data with one command. Give your models what they need.
🔗 github.com/NanoNets/docstran…
🔗 pypi.org/project/docstrange/
4
10
31
2,811
