Ishant Wankhede

Senior Data Scientist Walmart Labs | Mathematics and Computing, IIT Guwahati

[Ishant Wankhede @ishant30oct]

Careful with multiples of 2

[Andrej Karpathy @karpathy]

The most dramatic optimization to nanoGPT so far (~25% speedup) is to simply increase vocab size from 50257 to 50304 (nearest multiple of 64). This calculates added useless dimensions but goes down a different kernel path with much higher occupancy. Careful with your Powers of 2.

[Ishant Wankhede @ishant30oct]

@DeshKaaVikas Oh so this is how nandi hills looks like, that's why the craze!

[Ishant Wankhede @ishant30oct]

GPT 4 is coming!

[Ishant Wankhede @ishant30oct]

GPT-4 Will Have 100 Trillion Parameters — 500x the Size of GPT-3
https://t.co/X4bQphUnkq

[Ishant Wankhede @ishant30oct]

@CharmieKapoor @peakbengaluru Your kitchen garden is growing on pot biryanis'.

[Ishant Wankhede @ishant30oct]

@abhi1thakur :)

[Ishant Wankhede @ishant30oct]

Google has been using BERT in search since 2018, we only knew about it in 2019.

https://t.co/9Q3LPvN3Ah

[Ishant Wankhede @ishant30oct]

@VijaySundaram_ Question of 2011 answered in 2022.

[Ishant Wankhede @ishant30oct]

@alfcnz @stanley_h_chan Thankyou looks interesting

[Ishant Wankhede @ishant30oct]

@fchollet Rigt click refresh
Empty recycle bin
Delete cookies
Close all task manager

[Ishant Wankhede @ishant30oct]

Where does one begin reading about adversarial papers in NLP?

[Yanai Elazar @yanaiela]

Since you asked, here's my rant about adversarial examples papers in NLP:

[Ishant Wankhede @ishant30oct]

@xavierofficials Music director and lyricist during exams.

[Ishant Wankhede @ishant30oct]

@Tiwarihere Get well soon take care.

[Ishant Wankhede @ishant30oct]

@ranjodhd Managing PR for brands from Sarojini.

[Ishant Wankhede @ishant30oct]

@marktenenholtz Code fatigue is real. Would you share some of error analysis code recipe you would have developed over the years.

[Ishant Wankhede @ishant30oct]

Interesting. Visualising them as graph node made them all the more humane.

[Benji Weber @benjiweber]

Here's a team. Someone leaves.

We hire a replacement.

We get lucky and manage to find someone more skilled.

Looks like we're better off? [1/10]

[Ishant Wankhede @ishant30oct]

@Tiwarihere This is inspiring 🙌

[Ishant Wankhede @ishant30oct]

@Pal2PalKaShayar I don't go through any shit in life thinking it wouldn't make me any funnier. I haven't been wrong so far.

[Ishant Wankhede @ishant30oct]

@Pal2PalKaShayar @getplumhq Need heaby meta understanding to get it. Def 3 layer.

[Ishant Wankhede @ishant30oct]

GPT-4 Will Have 100 Trillion Parameters — 500x the Size of GPT-3
https://t.co/X4bQphUnkq

[Piyal Banik @PiyalBanik]

3 beginners level Machine Learning projects with code

- Regression
- Classification
- Clustering

🧵👇