All
Search
Images
Videos
Shorts
Maps
News
More
Shopping
Flights
Travel
Notebook
Report an inappropriate content
Please select one of the options below.
Not Relevant
Offensive
Adult
Child Sexual Abuse
Gaming PC
Solo Leveling
About DDR4 RAM
Tik Tok Para
PC
Battlefield 6
PC
Naptown Black Expo 2006
GeForce Now
GeForce
Assetto Corsa F1
PC
Componentes
Gaming
Editing
Grand Pro Circuit
Lave Roube Um Brainrot
Warranty Claim for NVIDIA
RTX 5090 Hire
Graphic Cards for
PC
Toledo Ohio James Willhauck
PC
Build
En Su Propia Trampa
GamePro
Bánh Mì Ram Ram
Antec Hcg850
Black Screen 10 Hours No Sound
Laurenzside Live Stream Bf
Apple Vision Pro 2025
Best Graphics Cards for Your Budget
NVIDIA
Trolls
Banh MI Ram Ram
PC Gaming
Setup
Miawaug
Length
All
Short (less than 5 minutes)
Medium (5-20 minutes)
Long (more than 20 minutes)
Date
All
Past 24 hours
Past week
Past month
Past year
Resolution
All
Lower than 360p
360p or higher
480p or higher
720p or higher
1080p or higher
Source
All
Dailymotion
Vimeo
Metacafe
Hulu
VEVO
Myspace
MTV
CBS
Fox
CNN
MSN
Price
All
Free
Paid
Clear filters
SafeSearch:
Moderate
Strict
Moderate (default)
Off
Filter
Gaming PC
Solo Leveling
About DDR4 RAM
Tik Tok Para
PC
Battlefield 6
PC
Naptown Black Expo 2006
GeForce Now
GeForce
Assetto Corsa F1
PC
Componentes
Gaming
Editing
Grand Pro Circuit
Lave Roube Um Brainrot
Warranty Claim for NVIDIA
RTX 5090 Hire
Graphic Cards for
PC
Toledo Ohio James Willhauck
PC
Build
En Su Propia Trampa
GamePro
Bánh Mì Ram Ram
Antec Hcg850
Black Screen 10 Hours No Sound
Laurenzside Live Stream Bf
Apple Vision Pro 2025
Best Graphics Cards for Your Budget
NVIDIA
Trolls
Banh MI Ram Ram
PC Gaming
Setup
Miawaug
0:13
You now convert any LLM into a faster one without retraining from scratch.NVIDIA just did this to their 30B model. Here's the trick:1. Duplicate the model into two copies2. Freeze one copy, it just reads the prompt and remembers context3. Train the other copy to write chunks of text at once instead of one word at a time4. Run them togetherThe frozen copy barely costs anything (it's already trained). The new copy only needed ~8% of the original training data to learn the new trick.Result: 2.4x fa
103.4K views
1 day ago
x.com
Lior Alexander
See more
More like this
Feedback