Student at @NITRourkela || Not a "x" developer || I write bugs🐛
Rourkela, Odisha, India
Joined January 2022
- Tweets672
- Following194
- Followers123
- Likes508
Pinned Tweet
🤔 Curious about who's behind the scenes crafting all things technical, websites, and digital magic during @hacknitr? Give them a laptop and they are ready to work anywhere. Ladies and gentlemen "The #hacknitr tech team"🤌
Glad to lead the team💪
#hackathon #tech #GoogleDeveloper
Does the language you calibrate llama.cpp quantization on matter for Indic languages? I ran a small pilot on Gemma 3 4B and Qwen3-4B: KL divergence vs F16 on FLORES+, MILU accuracy, and translation chrF++, with English vs Indic importance-matrix calibration at equal file size.
3
51
Built on llama.cpp (@ggerganov) and AI4Bharat's MILU (@ai4bharat), with FLORES+ for the parallel text. I would value corrections and model suggestions.
13
Do I have this weird habit or others also do this of linking my memories with songs.
Like I heard this song Ranjheya Ve when I visited Mysore, whenever I hear it again, magic, and I am in Mysore again😄
34
What should I read next? Any suggestions? Or how can I do some implementation on this?
Finished reading Inference Engineering by @philipkiely. To make it stick, I built a visual mind map for each chapter, plus figures you can play with: GPU memory, speculative decoding, autoscaling and more.
All here: ashutoshrath.me/notes/infere…
21
Finished reading Inference Engineering by @philipkiely. To make it stick, I built a visual mind map for each chapter, plus figures you can play with: GPU memory, speculative decoding, autoscaling and more.
All here: ashutoshrath.me/notes/infere…
Season 1 of The Fundamentals of Business by Michael Scott is out. Over 1 billion copies sold.
More than the Bible. It is thick, it is hard to put down, and you will want to finish in one sitting. That's what she said haaaa.
msbof.vercel.app/
22
sys1 scoreboard, all on held-out data:
phishing --> 100%
own domain --> 97.5%
typed decisions --> 73.4%
Laya scores 76.6% on that last one with 421M params. I'm at 68M, trained on a free Kaggle GPU.
Getting closerrrrrr........
38
Spent days trying to fix my model's phishing score.
58% accuracy. 17% recall - it answered "legitimate" to almost everything.
The fix turned out to be nothing I expected.
1
36
What I'm fixing next:
training on all 4 workflows (7,387 examples, was 2,887)
Platt scaling for yes/no: temperature fixes sharpness, not bias, and bias was my problem
bigger backbone with an MLM head, which my current one lacks
1
15
Everything reports a confidence interval now, because I'd been reading noise as progress. At n=500 it was +/-4 points - wider than most changes I was celebrating.
23M params. Laptop CPU. Free GPU.
github.com/ashutosh-rath02/s…
1
12