khoi
@khoidng.bsky.social
nuclear electrical engineer ⚡️☢️ | i wrote a lot of software past life chapter | learning more on advanced manufacturing, ML/RL, nuclear, and power research
Reposted by khoi
The current state of bird flu in America
December 31, 2024 at 5:39 AM
The current state of bird flu in America
Richland Public Library! Love how they have an entire section about nuclear history and nuclear physics. They also have records & old articles about the Hanford site as well.
Destroy this myth. Please share if you have visited and researched in a library this past year! RT And give a shout out to your favorite library you have ever visited.
December 21, 2024 at 10:27 AM
Richland Public Library! Love how they have an entire section about nuclear history and nuclear physics. They also have records & old articles about the Hanford site as well.
Reposted by khoi
Astonishing how many RL bottlenecks are resolved simply by “make simulator go fast”. What if we had prioritized engineering over algorithms years ago?
December 21, 2024 at 6:48 AM
Astonishing how many RL bottlenecks are resolved simply by “make simulator go fast”. What if we had prioritized engineering over algorithms years ago?
Reposted by khoi
"Maryland-based advanced reactor developer X-energy and its wholly-owned subsidiary Triso-X have chosen Ohio-based construction and engineering company Geiger Brothers to develop the site of a first-of-a-kind advanced nuclear fuel fabrication facility in Oak Ridge, Tennessee.
The facility […]
The facility […]
Original post on greennuclear.online
greennuclear.online
December 21, 2024 at 9:00 AM
"Maryland-based advanced reactor developer X-energy and its wholly-owned subsidiary Triso-X have chosen Ohio-based construction and engineering company Geiger Brothers to develop the site of a first-of-a-kind advanced nuclear fuel fabrication facility in Oak Ridge, Tennessee.
The facility […]
The facility […]
Reposted by khoi
Starting to develop a real fondness for the social media site preferred by scientists over the one preferred by arsonists.
November 27, 2024 at 10:10 PM
Starting to develop a real fondness for the social media site preferred by scientists over the one preferred by arsonists.
Reposted by khoi
Want to learn / teach RL?
Check out new book draft:
Reinforcement Learning - Foundations sites.google.com/view/rlfound...
W/ Shie Mannor & Yishay Mansour
This is a rigorous first course in RL, based on our teaching at TAU CS and Technion ECE.
Check out new book draft:
Reinforcement Learning - Foundations sites.google.com/view/rlfound...
W/ Shie Mannor & Yishay Mansour
This is a rigorous first course in RL, based on our teaching at TAU CS and Technion ECE.
November 25, 2024 at 12:08 PM
Want to learn / teach RL?
Check out new book draft:
Reinforcement Learning - Foundations sites.google.com/view/rlfound...
W/ Shie Mannor & Yishay Mansour
This is a rigorous first course in RL, based on our teaching at TAU CS and Technion ECE.
Check out new book draft:
Reinforcement Learning - Foundations sites.google.com/view/rlfound...
W/ Shie Mannor & Yishay Mansour
This is a rigorous first course in RL, based on our teaching at TAU CS and Technion ECE.
Reposted by khoi
This is actually a really cool interview, wish I could read it properly (here's a google translate link: mp-weixin-qq-com.translate.goog/s/r9zZaEgqAa...)
November 25, 2024 at 5:57 PM
This is actually a really cool interview, wish I could read it properly (here's a google translate link: mp-weixin-qq-com.translate.goog/s/r9zZaEgqAa...)
Reposted by khoi
Trying to build a "books you must read" list for my lab that everyone gets when they enter. Right now its:
- Sutton and Barto
- The Structure of Scientific Revolutions
- Strunk and White
- Maybe "Prediction, Learning, and Games", TBD
Kinda curious what's missing in an RL / science curriculum
- Sutton and Barto
- The Structure of Scientific Revolutions
- Strunk and White
- Maybe "Prediction, Learning, and Games", TBD
Kinda curious what's missing in an RL / science curriculum
November 25, 2024 at 5:43 PM
Trying to build a "books you must read" list for my lab that everyone gets when they enter. Right now its:
- Sutton and Barto
- The Structure of Scientific Revolutions
- Strunk and White
- Maybe "Prediction, Learning, and Games", TBD
Kinda curious what's missing in an RL / science curriculum
- Sutton and Barto
- The Structure of Scientific Revolutions
- Strunk and White
- Maybe "Prediction, Learning, and Games", TBD
Kinda curious what's missing in an RL / science curriculum
Reposted by khoi
Everyone deserves to know about Styropyro. Or Drake
"I'm a big believer that eBay is one of the greatest things to come from the internet age. I mean there's always so much interesting for sale.
Where else am I gonna get my Soviet military surplus or fusion laser optics?"
youtu.be/DMVWW-bmKwQ
"I'm a big believer that eBay is one of the greatest things to come from the internet age. I mean there's always so much interesting for sale.
Where else am I gonna get my Soviet military surplus or fusion laser optics?"
youtu.be/DMVWW-bmKwQ
November 24, 2024 at 3:37 PM
Everyone deserves to know about Styropyro. Or Drake
"I'm a big believer that eBay is one of the greatest things to come from the internet age. I mean there's always so much interesting for sale.
Where else am I gonna get my Soviet military surplus or fusion laser optics?"
youtu.be/DMVWW-bmKwQ
"I'm a big believer that eBay is one of the greatest things to come from the internet age. I mean there's always so much interesting for sale.
Where else am I gonna get my Soviet military surplus or fusion laser optics?"
youtu.be/DMVWW-bmKwQ
Reposted by khoi
Everything's fine, I just binge-bought eleven books in one night
November 23, 2024 at 5:00 PM
Everything's fine, I just binge-bought eleven books in one night
Reposted by khoi
I want to believe this is real boingboing.net/2024/11/21/f...
Factory robot convinces 12 other robots to go on strike
A robotics company in Shanghai wanted to see how effective its persuasive AI programming could be.
boingboing.net
November 24, 2024 at 12:51 AM
I want to believe this is real boingboing.net/2024/11/21/f...
Reposted by khoi
One thing I find interesting about the echo chamber complaint is that an echo chamber just won a presidential election.
Put aside the convo of whether it’s “good” or “bad” for a moment. The far right has devoted enormous resources + time to building echo chambers & loathes spaces they don’t control
Put aside the convo of whether it’s “good” or “bad” for a moment. The far right has devoted enormous resources + time to building echo chambers & loathes spaces they don’t control
The "bluesky is an echo chamber" think pieces that keep popping up really tickle me because they show how certain people genuinely believe social media should be for debating and arguing and not talking about the things that make you happy and sharing art with the world.
November 23, 2024 at 1:29 AM
One thing I find interesting about the echo chamber complaint is that an echo chamber just won a presidential election.
Put aside the convo of whether it’s “good” or “bad” for a moment. The far right has devoted enormous resources + time to building echo chambers & loathes spaces they don’t control
Put aside the convo of whether it’s “good” or “bad” for a moment. The far right has devoted enormous resources + time to building echo chambers & loathes spaces they don’t control
Reposted by khoi
These seem like some very common (statistical) sense recommendations for eval. Still the title of the paper is somewhat confusing as if these ideas are new. I would’ve suggested “Bringing a statistical approach…”
www.anthropic.com/research/sta...
www.anthropic.com/research/sta...
A statistical approach to model evaluations
A research paper from Anthropic on how to apply statistics to improve language model evaluations
www.anthropic.com
November 22, 2024 at 10:39 AM
These seem like some very common (statistical) sense recommendations for eval. Still the title of the paper is somewhat confusing as if these ideas are new. I would’ve suggested “Bringing a statistical approach…”
www.anthropic.com/research/sta...
www.anthropic.com/research/sta...
Reposted by khoi
I really appreciate @whatisnuclear.com giving a great presentation to my team on nuclear development history! He has a great page on the subject. Check it out. whatisnuclear.com/reactor-hist...
Nuclear Reactor Development History
Learn about the wide variety of nuclear reactor development programs that have happened up to about 1970
whatisnuclear.com
November 23, 2024 at 12:59 AM
I really appreciate @whatisnuclear.com giving a great presentation to my team on nuclear development history! He has a great page on the subject. Check it out. whatisnuclear.com/reactor-hist...
Reposted by khoi
If Nature says so, it must be true.
‘A place of joy’: why scientists are joining the rush to Bluesky
Researchers say the social-media platform — an alternative to X — offers more control over the content they see and the people they engage with.
www.nature.com
November 22, 2024 at 12:12 PM
If Nature says so, it must be true.
Reposted by khoi
A timely paper exploring ways academics can pretrain larger models than they think, e.g. by trading time against GPU count.
Since the title is misleading, let me also say: US academics do not need $100k for this. They used 2,000 GPU hours in this paper; NSF will give you that. #MLSky
Since the title is misleading, let me also say: US academics do not need $100k for this. They used 2,000 GPU hours in this paper; NSF will give you that. #MLSky
$100K or 100 Days: Trade-offs when Pre-Training with Academic Resources
Pre-training is notoriously compute-intensive and academic researchers are notoriously under-resourced. It is, therefore, commonly assumed that academics can't pre-train models. In this paper, we seek...
arxiv.org
November 23, 2024 at 1:50 PM
A timely paper exploring ways academics can pretrain larger models than they think, e.g. by trading time against GPU count.
Since the title is misleading, let me also say: US academics do not need $100k for this. They used 2,000 GPU hours in this paper; NSF will give you that. #MLSky
Since the title is misleading, let me also say: US academics do not need $100k for this. They used 2,000 GPU hours in this paper; NSF will give you that. #MLSky
Reposted by khoi
Just realized BlueSky allows sharing valuable stuff cause it doesn't punish links. 🤩
Let's start with "What are embeddings" by @vickiboykis.com
The book is a great summary of embeddings, from history to modern approaches.
The best part: it's free.
Link: vickiboykis.com/what_are_emb...
Let's start with "What are embeddings" by @vickiboykis.com
The book is a great summary of embeddings, from history to modern approaches.
The best part: it's free.
Link: vickiboykis.com/what_are_emb...
November 22, 2024 at 11:13 AM
Just realized BlueSky allows sharing valuable stuff cause it doesn't punish links. 🤩
Let's start with "What are embeddings" by @vickiboykis.com
The book is a great summary of embeddings, from history to modern approaches.
The best part: it's free.
Link: vickiboykis.com/what_are_emb...
Let's start with "What are embeddings" by @vickiboykis.com
The book is a great summary of embeddings, from history to modern approaches.
The best part: it's free.
Link: vickiboykis.com/what_are_emb...
Reposted by khoi
🌶️(?) take: Agents are somehow hot right because people realized that LLM output can be interpreted as a DSL which directs side effects in the world (e.g. tool calls) rather than just returning text in a chat/autocomplete sense. What are the open challenges? A 🧵... [1/11]
November 19, 2024 at 9:32 AM
🌶️(?) take: Agents are somehow hot right because people realized that LLM output can be interpreted as a DSL which directs side effects in the world (e.g. tool calls) rather than just returning text in a chat/autocomplete sense. What are the open challenges? A 🧵... [1/11]
Reposted by khoi
An RL Library in C++ that seems pretty fast github.com/rl-tools/rl-..., pretty amazing that someone managed to code all of that stuff in C++ and that it's so efficient (it can even train models in an apple watch!)
GitHub - rl-tools/rl-tools: The Fastest Deep Reinforcement Learning Library
The Fastest Deep Reinforcement Learning Library. Contribute to rl-tools/rl-tools development by creating an account on GitHub.
github.com
November 20, 2024 at 6:36 PM
An RL Library in C++ that seems pretty fast github.com/rl-tools/rl-..., pretty amazing that someone managed to code all of that stuff in C++ and that it's so efficient (it can even train models in an apple watch!)
Reposted by khoi
The right better adopt this position real quick if it wants the US to have any hope of winning a military conflict in the 21st century.
Do you want your drones using American made firmware or not?
Do you want your drones using American made firmware or not?
November 21, 2024 at 3:12 AM
The right better adopt this position real quick if it wants the US to have any hope of winning a military conflict in the 21st century.
Do you want your drones using American made firmware or not?
Do you want your drones using American made firmware or not?
Reposted by khoi
Nuclear Weapons rated pallet jacks are pretty hefty!
November 21, 2024 at 5:24 AM
Nuclear Weapons rated pallet jacks are pretty hefty!
Reposted by khoi
Reposted by khoi
64th #Top500 is out!
o #ElCapitan at LLNL is the new #1
o We now have 3 #Exascale systems officially
o #HPC6 at Eni is new at #5
o Tuolumne, a smaller version of #ElCapitan, rounds up the #Top10
top500.org/news/el-capi...
#HPC #AI #SC24
o #ElCapitan at LLNL is the new #1
o We now have 3 #Exascale systems officially
o #HPC6 at Eni is new at #5
o Tuolumne, a smaller version of #ElCapitan, rounds up the #Top10
top500.org/news/el-capi...
#HPC #AI #SC24
November 18, 2024 at 5:53 PM
64th #Top500 is out!
o #ElCapitan at LLNL is the new #1
o We now have 3 #Exascale systems officially
o #HPC6 at Eni is new at #5
o Tuolumne, a smaller version of #ElCapitan, rounds up the #Top10
top500.org/news/el-capi...
#HPC #AI #SC24
o #ElCapitan at LLNL is the new #1
o We now have 3 #Exascale systems officially
o #HPC6 at Eni is new at #5
o Tuolumne, a smaller version of #ElCapitan, rounds up the #Top10
top500.org/news/el-capi...
#HPC #AI #SC24
Reposted by khoi
November 19, 2024 at 12:57 AM
Reposted by khoi
David T. Frazier, Ryan Kelly, Christopher Drovandi, David J. Warne
The Statistical Accuracy of Neural Posterior and Likelihood Estimation
https://arxiv.org/abs/2411.12068
The Statistical Accuracy of Neural Posterior and Likelihood Estimation
https://arxiv.org/abs/2411.12068
November 20, 2024 at 5:01 AM
David T. Frazier, Ryan Kelly, Christopher Drovandi, David J. Warne
The Statistical Accuracy of Neural Posterior and Likelihood Estimation
https://arxiv.org/abs/2411.12068
The Statistical Accuracy of Neural Posterior and Likelihood Estimation
https://arxiv.org/abs/2411.12068