HK
Heykuki News
Top
New
Best
Ask
Show
Jobs
Toggle theme
Top
New
Best
Ask
Show
Jobs
Request
151.
▲
Story of development and inspiration behind the "attention" operator
twitter.com
discuss
2 years ago
aberoham
2 points
152.
▲
The Weirdness of LLM Tokenization
twitter.com
discuss
2 years ago
tosh
2 points
153.
▲
Consider Being a Labeler for an LLM
twitter.com
discuss
2 years ago
tosh
2 points
154.
▲
Scheduling Workloads to Run on Humans
twitter.com
discuss
2 years ago
tosh
2 points
155.
▲
Llm.c is only 2X slower than PyTorch (fp32, forward pass)
twitter.com
discuss
2 years ago
tosh
2 points
156.
▲
A Deepdive into the Gemma Tokenizer
twitter.com
discuss
2 years ago
tosh
2 points
157.
▲
Andrej Karpathy on Apple Vision Pro
twitter.com
discuss
2 years ago
SerCe
2 points
158.
▲
On the "Hallucination Problem"
twitter.com
discuss
3 years ago
tosh
2 points
159.
▲
LLM Bootcamp – Spring 2023
twitter.com
discuss
3 years ago
andromaton
2 points
160.
▲
The Chinchilla Trap
twitter.com
discuss
3 years ago
debdut
2 points
161.
▲
“With a single readable 600 line main.py, bunch of nice tricks” – Karapathy
twitter.com
discuss
3 years ago
textread
2 points
162.
▲
The hottest new programming language is English
twitter.com
discuss
3 years ago
tafath
2 points
163.
▲
GPT is all you need for back end
twitter.com
discuss
3 years ago
WithinReason
2 points
164.
▲
Andrej Karpathy has been on a 4 month sabbatical
twitter.com
discuss
4 years ago
sarthakjshetty
2 points
165.
▲
Math.sqrt vs. Numpy.sqrt vs x ** 0.5
twitter.com
discuss
5 years ago
marinesebastian
2 points
166.
▲
Browsing the Web, 2021
twitter.com
discuss
5 years ago
stanislavb
2 points
167.
▲
Browsing the Web in 2021
twitter.com
discuss
5 years ago
nafizh
2 points
168.
▲
TenserFlow mentioned in ~6% papers in last 6 years, highest
twitter.com
discuss
8 years ago
gajju3588
2 points
169.
▲
Andrej Karpathy forced to take down CS231n videos
twitter.com
discuss
10 years ago
Gimpei
2 points
170.
▲
I had ~30 direct reports and didn't do 1on1s at Tesla and imo it was great
twitter.com
2 comments
2 years ago
tosh
1 points
171.
▲
Software 1.0 easily automates what you can specify, 2.0 what you can verify
twitter.com
1 comment
7 months ago
speckx
1 points
172.
▲
Andrej Karpathy on improved Tesla FSD performance under HW4
twitter.com
1 comment
7 months ago
maxutility
1 points
173.
▲
Agency > Intelligence (Karpathy)
twitter.com
1 comment
8 months ago
andsoitis
1 points
174.
▲
What the founding fathers would have thought about today's America
twitter.com
1 comment
2 years ago
tosh
1 points
175.
▲
Highly Bespoke Software
twitter.com
discuss
4 months ago
tosh
1 points
176.
▲
DeepWiki and Increasing Malleability of Software
twitter.com
discuss
4 months ago
sabareesh
1 points
177.
▲
Train and inference GPT in 243 lines of pure, dependency-free Python
twitter.com
discuss
4 months ago
tosh
1 points
178.
▲
DeepWiki and Increasing Malleability of Software
twitter.com
discuss
4 months ago
ipnon
1 points
179.
▲
Beating GPT-2 for <<$100: the nanochat journey
twitter.com
discuss
5 months ago
rzk
1 points
180.
▲
I'm being accused of overhyping (Karpathy)
twitter.com
discuss
5 months ago
baxtr
1 points
More