Nusret Ozates
@nusretozates.bsky.social
ML/Ops Engineer | MS at Koc University studying Computer Vision in Pathology | Working on NLP in the industry | Google Cloud Certified - Professional ML Engineer
I wanted to try GPT-5 by making it create a refactoring plan for one of the modules in my thesis work. The plan looked fine and then I said okay start to implement it. Now I have no idea how that module works or if it is working lol... I will reverse some commits probably
August 8, 2025 at 5:14 PM
I wanted to try GPT-5 by making it create a refactoring plan for one of the modules in my thesis work. The plan looked fine and then I said okay start to implement it. Now I have no idea how that module works or if it is working lol... I will reverse some commits probably
I hate these questions
July 31, 2025 at 3:23 PM
I hate these questions
My biggest mistake in 2024 was listening a guy who said to buy eth as he got lots of money and it will go much higher. And my biggest gratitude is realizing it is a mistake (after losing some money) and running away earlier 😂
April 7, 2025 at 2:58 PM
My biggest mistake in 2024 was listening a guy who said to buy eth as he got lots of money and it will go much higher. And my biggest gratitude is realizing it is a mistake (after losing some money) and running away earlier 😂
PanNuke is a flawed dataset and having a good score on it actually a bad thing? For example, left image is an example from the dataset, the middle is ground truth and right is my model's prediction. According to a doctor friend my model is right and GT is wrong. How can a model learn from this?
March 28, 2025 at 10:48 AM
PanNuke is a flawed dataset and having a good score on it actually a bad thing? For example, left image is an example from the dataset, the middle is ground truth and right is my model's prediction. According to a doctor friend my model is right and GT is wrong. How can a model learn from this?
Self-supervised Swin Transformer on Pathology domain when? Recent article shows that Imagenet Swin > UNI 2 and other pathology foundation models with vanilla ViT for cell segmentation
March 25, 2025 at 8:40 PM
Self-supervised Swin Transformer on Pathology domain when? Recent article shows that Imagenet Swin > UNI 2 and other pathology foundation models with vanilla ViT for cell segmentation
In my MSc study in medical image processing (with CV) I feel like I'm missing a lot in LLM side but on the other hand it feels like nothing really "new" happens except bigger data and bigger models. Though I think I must read all deepseek papers
March 10, 2025 at 4:35 PM
In my MSc study in medical image processing (with CV) I feel like I'm missing a lot in LLM side but on the other hand it feels like nothing really "new" happens except bigger data and bigger models. Though I think I must read all deepseek papers
I will give a training about ML on the GCP in 5 mins 🎤
March 8, 2025 at 5:39 PM
I will give a training about ML on the GCP in 5 mins 🎤
So it turns out using class weight for segmentation is not good especially if one of your class weight is 0.20 while there is also a class with weight 240
March 2, 2025 at 12:40 AM
So it turns out using class weight for segmentation is not good especially if one of your class weight is 0.20 while there is also a class with weight 240
Twitter is toxic and bluesky seems empty...
February 24, 2025 at 12:00 PM
Twitter is toxic and bluesky seems empty...
Hey everyone, once again I proved that I'm a pro about GCP and ML *mic drop* hehe😁 This certificate basically means I know my way around GCP when a business problem needs to be solved using machine learning on the Google Cloud!
February 22, 2025 at 7:48 AM
Hey everyone, once again I proved that I'm a pro about GCP and ML *mic drop* hehe😁 This certificate basically means I know my way around GCP when a business problem needs to be solved using machine learning on the Google Cloud!
It turns out you have to do that to have a better bluesky experience
February 10, 2025 at 8:58 PM
It turns out you have to do that to have a better bluesky experience
I have things to do, but I've suddenly been struck by inspiration, and since the well-known work-avoidance mechanism has kicked in, I'm going to write down my thoughts on test-time compute, shared decoders, and reasoning. A theoretical and lengthy piece is coming.
medium.com/@m.nusret.oz...
medium.com/@m.nusret.oz...
Test-Time Compute, Reasoning and Human Brain
I have lots of things to do, but I’ve suddenly been struck by inspiration, and since the well-known work-avoidance mechanism has kicked in…
medium.com
February 7, 2025 at 8:17 PM
I have things to do, but I've suddenly been struck by inspiration, and since the well-known work-avoidance mechanism has kicked in, I'm going to write down my thoughts on test-time compute, shared decoders, and reasoning. A theoretical and lengthy piece is coming.
medium.com/@m.nusret.oz...
medium.com/@m.nusret.oz...
lol I was wrong back to the reading and thinking and experimenting again
I think I finally found my thesis topic 🎉🎊 Just need a little bit more experiment and some discussions with my advisor now
February 2, 2025 at 10:17 AM
lol I was wrong back to the reading and thinking and experimenting again
Choose your fighter👇 Additional information:
- All models are for the same task
- There is also a training py for all but I forget to add
- All models needs to use a dataset . py and maybe other scripts, so think about where to put it and how it would change the structure
- All models are for the same task
- There is also a training py for all but I forget to add
- All models needs to use a dataset . py and maybe other scripts, so think about where to put it and how it would change the structure
January 27, 2025 at 4:27 PM
Choose your fighter👇 Additional information:
- All models are for the same task
- There is also a training py for all but I forget to add
- All models needs to use a dataset . py and maybe other scripts, so think about where to put it and how it would change the structure
- All models are for the same task
- There is also a training py for all but I forget to add
- All models needs to use a dataset . py and maybe other scripts, so think about where to put it and how it would change the structure
It is funny to see binary opening/closing and watershed is still very useful for segmentation when combined with deep learning. Why funny? Because when I first learn about them I thought they were the things from ancient history and not used anymore
January 12, 2025 at 6:11 AM
It is funny to see binary opening/closing and watershed is still very useful for segmentation when combined with deep learning. Why funny? Because when I first learn about them I thought they were the things from ancient history and not used anymore
I think I finally found my thesis topic 🎉🎊 Just need a little bit more experiment and some discussions with my advisor now
January 7, 2025 at 8:45 AM
I think I finally found my thesis topic 🎉🎊 Just need a little bit more experiment and some discussions with my advisor now
Happy new year everyone 🎊
December 31, 2024 at 4:12 PM
Happy new year everyone 🎊
Reposted by Nusret Ozates
You’re still arguing about tabs vs. spaces? May I present…
December 25, 2024 at 6:37 PM
You’re still arguing about tabs vs. spaces? May I present…
It seems like bigger vision transformer models need extra tokens (other than cls) to store more global information. Otherwise they remove local information from some patches and use them as global context holders. "Vision Transformers Need Registers" by Meta (paper link below)
December 25, 2024 at 9:51 AM
It seems like bigger vision transformer models need extra tokens (other than cls) to store more global information. Otherwise they remove local information from some patches and use them as global context holders. "Vision Transformers Need Registers" by Meta (paper link below)
So the drop in coin prices in the last 2 days was a strong reminder for me that I need to find a thesis topic and finish is very, very, very fast and find a job otherwise I will be broke much faster than I planned 😂
December 18, 2024 at 9:30 PM
So the drop in coin prices in the last 2 days was a strong reminder for me that I need to find a thesis topic and finish is very, very, very fast and find a job otherwise I will be broke much faster than I planned 😂
Reposted by Nusret Ozates
Three must read papers for PhD students. #scisky #PhD #science #research #academicsky
1. The importance of stupidity in scientific research
Open Access
journals.biologists.com/jcs/article/...
1. The importance of stupidity in scientific research
Open Access
journals.biologists.com/jcs/article/...
November 24, 2024 at 1:54 PM
Three must read papers for PhD students. #scisky #PhD #science #research #academicsky
1. The importance of stupidity in scientific research
Open Access
journals.biologists.com/jcs/article/...
1. The importance of stupidity in scientific research
Open Access
journals.biologists.com/jcs/article/...
Reposted by Nusret Ozates
Missing AutoML 24? We feel the same! 😢 But don’t worry - you can relive the highlights:
🎥 Watch the talks: www.youtube.com/@automl_conf...
📚 Explore the proceedings: proceedings.mlr.press/v256/
And start planning for AutoML 25! 2025.automl.cc/call-for-pap...
#AutoML25 #AutoMLConf #NYC #AutoML
🎥 Watch the talks: www.youtube.com/@automl_conf...
📚 Explore the proceedings: proceedings.mlr.press/v256/
And start planning for AutoML 25! 2025.automl.cc/call-for-pap...
#AutoML25 #AutoMLConf #NYC #AutoML
December 10, 2024 at 11:12 AM
Missing AutoML 24? We feel the same! 😢 But don’t worry - you can relive the highlights:
🎥 Watch the talks: www.youtube.com/@automl_conf...
📚 Explore the proceedings: proceedings.mlr.press/v256/
And start planning for AutoML 25! 2025.automl.cc/call-for-pap...
#AutoML25 #AutoMLConf #NYC #AutoML
🎥 Watch the talks: www.youtube.com/@automl_conf...
📚 Explore the proceedings: proceedings.mlr.press/v256/
And start planning for AutoML 25! 2025.automl.cc/call-for-pap...
#AutoML25 #AutoMLConf #NYC #AutoML
I've tried Gemini 2 flash and wanted it to act like a math professor in Stanford. When I ask questions, it "quotes" my words and I kind of feel like it looks down on me 😂
December 13, 2024 at 3:39 PM
I've tried Gemini 2 flash and wanted it to act like a math professor in Stanford. When I ask questions, it "quotes" my words and I kind of feel like it looks down on me 😂
Me: Professional Google Certified Superhero Machine Learning MLOps Engineer
My Model:
My Model:
December 5, 2024 at 9:19 PM
Me: Professional Google Certified Superhero Machine Learning MLOps Engineer
My Model:
My Model:
When I cluster the features from the UNI model,
2 clusters: Foreground background
3 clusters: Border information added
4 clusters: Parts where the model isn't sure if it is a border or background added
Now I need to find an original way to use this information for designing a segmentation decoder
2 clusters: Foreground background
3 clusters: Border information added
4 clusters: Parts where the model isn't sure if it is a border or background added
Now I need to find an original way to use this information for designing a segmentation decoder
December 4, 2024 at 3:47 PM
When I cluster the features from the UNI model,
2 clusters: Foreground background
3 clusters: Border information added
4 clusters: Parts where the model isn't sure if it is a border or background added
Now I need to find an original way to use this information for designing a segmentation decoder
2 clusters: Foreground background
3 clusters: Border information added
4 clusters: Parts where the model isn't sure if it is a border or background added
Now I need to find an original way to use this information for designing a segmentation decoder