Walter Hernandez
@walterhernandez.bsky.social
Researching and learning about AI (mainly ML and NLP) and DLT (mainly AMMs and stablecoins) at UCL and @exponentialscience.bsky.social
Pinned
I could not find an starter pack for researchers in Distributed Ledger Technology and related topics, so I created one:
go.bsky.app/AtpHdQh
Let me know for any researcher in Bluesky that I may be missing in the starter pack
go.bsky.app/AtpHdQh
Let me know for any researcher in Bluesky that I may be missing in the starter pack
Reposted by Walter Hernandez
Nice to see another fully open, multimodal LM released! Good license, training code, pretraining data, all here.
LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training
Slowly, the community is growing.
arxiv.org/abs/2509.236...
LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training
Slowly, the community is growing.
arxiv.org/abs/2509.236...
September 30, 2025 at 4:03 PM
Nice to see another fully open, multimodal LM released! Good license, training code, pretraining data, all here.
LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training
Slowly, the community is growing.
arxiv.org/abs/2509.236...
LLaVA-OneVision-1.5: Fully Open Framework for Democratized Multimodal Training
Slowly, the community is growing.
arxiv.org/abs/2509.236...
Reposted by Walter Hernandez
It's been three years now of nothing by LLMs in every NLP conference (and a large chunk of the ML venues too).
LLMs are fascinating, but is there really nothing else worth researching in NLP anymore?
LLMs are fascinating, but is there really nothing else worth researching in NLP anymore?
May 17, 2025 at 6:42 PM
It's been three years now of nothing by LLMs in every NLP conference (and a large chunk of the ML venues too).
LLMs are fascinating, but is there really nothing else worth researching in NLP anymore?
LLMs are fascinating, but is there really nothing else worth researching in NLP anymore?
Reposted by Walter Hernandez
Only a quarter of AI initiatives have delivered the expected return on investment, according to a survey of 2,000 CEOs.
Companies are struggling to get value from #GenAI. Most of the adoption of the technology is based on FOMO.
#AIEthics
www.theregister.com/2025/05/06/i...
Companies are struggling to get value from #GenAI. Most of the adoption of the technology is based on FOMO.
#AIEthics
www.theregister.com/2025/05/06/i...
Most AI spending driven by FOMO, not ROI, CEOs tell IBM
: Just 1 in 4 bets paying off so far
www.theregister.com
May 10, 2025 at 9:17 AM
Only a quarter of AI initiatives have delivered the expected return on investment, according to a survey of 2,000 CEOs.
Companies are struggling to get value from #GenAI. Most of the adoption of the technology is based on FOMO.
#AIEthics
www.theregister.com/2025/05/06/i...
Companies are struggling to get value from #GenAI. Most of the adoption of the technology is based on FOMO.
#AIEthics
www.theregister.com/2025/05/06/i...
Reposted by Walter Hernandez
"Science is an investment.
We will put forward a new 500 million package for 2025-2027 to support the best and the brightest researchers and scientists from Europe and around the world."
— President @vonderleyen.ec.europa.eu at the ‘Choose Europe for Science' event at La Sorbonne 🇫🇷
We will put forward a new 500 million package for 2025-2027 to support the best and the brightest researchers and scientists from Europe and around the world."
— President @vonderleyen.ec.europa.eu at the ‘Choose Europe for Science' event at La Sorbonne 🇫🇷
May 5, 2025 at 10:16 AM
"Science is an investment.
We will put forward a new 500 million package for 2025-2027 to support the best and the brightest researchers and scientists from Europe and around the world."
— President @vonderleyen.ec.europa.eu at the ‘Choose Europe for Science' event at La Sorbonne 🇫🇷
We will put forward a new 500 million package for 2025-2027 to support the best and the brightest researchers and scientists from Europe and around the world."
— President @vonderleyen.ec.europa.eu at the ‘Choose Europe for Science' event at La Sorbonne 🇫🇷
Reposted by Walter Hernandez
A new paper, "Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?", has people reconsidering if the RL we're hearing about really works.
It shows RL elicits from the models, but as we get better verifiers we may not need to rely on RL as much.
Good read.
It shows RL elicits from the models, but as we get better verifiers we may not need to rely on RL as much.
Good read.
April 21, 2025 at 4:17 PM
A new paper, "Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?", has people reconsidering if the RL we're hearing about really works.
It shows RL elicits from the models, but as we get better verifiers we may not need to rely on RL as much.
Good read.
It shows RL elicits from the models, but as we get better verifiers we may not need to rely on RL as much.
Good read.
Reposted by Walter Hernandez
Multi-node, multi-GPU training is pretty easy with torchrun, just a few extra lines of code. Putting this out there into the world so people don't shy away from it
April 21, 2025 at 1:11 AM
Multi-node, multi-GPU training is pretty easy with torchrun, just a few extra lines of code. Putting this out there into the world so people don't shy away from it
Reposted by Walter Hernandez
Wrote up some notes on the o3/o4-mini system card, including my frustration at "sandbagging" joining the ever-growing collection of AI terminology with more than one competing definition https://simonwillison.net/2025/Apr/21/openai-o3-and-o4-mini-system-card/
April 21, 2025 at 7:19 PM
Wrote up some notes on the o3/o4-mini system card, including my frustration at "sandbagging" joining the ever-growing collection of AI terminology with more than one competing definition https://simonwillison.net/2025/Apr/21/openai-o3-and-o4-mini-system-card/
Reposted by Walter Hernandez
A new study has found that the universe might be spinning. What does that even mean? Let’s have a look.
www.youtube.com/watch?v=Gm5n...
www.youtube.com/watch?v=Gm5n...
The Entire Universe Seems to Spin, New Data Reveal
YouTube video by Sabine Hossenfelder
www.youtube.com
April 12, 2025 at 3:33 PM
A new study has found that the universe might be spinning. What does that even mean? Let’s have a look.
www.youtube.com/watch?v=Gm5n...
www.youtube.com/watch?v=Gm5n...
Reposted by Walter Hernandez
Time to remind ourselves of some observations about how trade appears to help stabilize alliances and prevent international conflict www.gsb.stanford.edu/insights/mat...
Matthew O. Jackson: Can Trade Prevent War?
www.gsb.stanford.edu
April 3, 2025 at 4:40 PM
Time to remind ourselves of some observations about how trade appears to help stabilize alliances and prevent international conflict www.gsb.stanford.edu/insights/mat...
Reposted by Walter Hernandez
“Can you draw a photorealistic beach with no elephants?”
March 26, 2025 at 9:09 AM
“Can you draw a photorealistic beach with no elephants?”
Reposted by Walter Hernandez
In my latest column for Science magazine, I discuss recent AI "reasoning" models -- how it works, to what extent it captures "genuine" reasoning processes, and what's needed to answer such questions.
www.science.org/doi/10.1126/...
www.science.org/doi/10.1126/...
Artificial intelligence learns to reason
Julia has two sisters and one brother. How many sisters does her brother Martin have?Solving this tiny puzzle requires a bit of thinking. You might mentally picture the family of three girls and one b...
www.science.org
March 20, 2025 at 6:40 PM
In my latest column for Science magazine, I discuss recent AI "reasoning" models -- how it works, to what extent it captures "genuine" reasoning processes, and what's needed to answer such questions.
www.science.org/doi/10.1126/...
www.science.org/doi/10.1126/...
Reposted by Walter Hernandez
This is absurdly great, but I haven't read a single news article about it. A fully open source, offline-first alternative to Notion that's a collab between the French and German governments because they want to host docs securely and on their own terms. THIS is what Europe should be doing.
Docs
Docs: Your new companion to collaborate on documents efficiently, intuitively, and securely.
docs.numerique.gouv.fr
March 16, 2025 at 11:03 PM
This is absurdly great, but I haven't read a single news article about it. A fully open source, offline-first alternative to Notion that's a collab between the French and German governments because they want to host docs securely and on their own terms. THIS is what Europe should be doing.
Reposted by Walter Hernandez
Oxford researchers have helped develop WildPose, a groundbreaking system using LiDAR & high-speed imaging to track wildlife in 3D from over 100m away. Capturing fine details like a lion’s breathing, it offers new insights into animal movement without invasive methods. www.cs.ox.ac.uk/news/2430-fu...
March 17, 2025 at 11:07 AM
Oxford researchers have helped develop WildPose, a groundbreaking system using LiDAR & high-speed imaging to track wildlife in 3D from over 100m away. Capturing fine details like a lion’s breathing, it offers new insights into animal movement without invasive methods. www.cs.ox.ac.uk/news/2430-fu...
Reposted by Walter Hernandez
Is everyone now okay with using the term "thinking" to describe what LLM "reasoning" models do? And to call their outputs "thoughts"?
From OpenAI blog posts:
From OpenAI blog posts:
March 14, 2025 at 9:12 PM
Is everyone now okay with using the term "thinking" to describe what LLM "reasoning" models do? And to call their outputs "thoughts"?
From OpenAI blog posts:
From OpenAI blog posts:
Reposted by Walter Hernandez
Reposted by Walter Hernandez
A lot of people lately are conflating novelty with unfamiliarity.
It explains all the responses of "this isn't new" to explanatory pieces which aren't claiming to be presenting new information. They're just trying to increase awareness.
It explains all the responses of "this isn't new" to explanatory pieces which aren't claiming to be presenting new information. They're just trying to increase awareness.
March 14, 2025 at 3:42 PM
A lot of people lately are conflating novelty with unfamiliarity.
It explains all the responses of "this isn't new" to explanatory pieces which aren't claiming to be presenting new information. They're just trying to increase awareness.
It explains all the responses of "this isn't new" to explanatory pieces which aren't claiming to be presenting new information. They're just trying to increase awareness.
Reposted by Walter Hernandez
So one good thing that seems to be happening right now is that a new end-to-end encryption standard "MLS" seems to be gaining a lot of momentum. Like, a lot.
And from what I understand this is an important step there as well, because RCS' encryption is MLS. Security folks correct me if I'm wrong
And from what I understand this is an important step there as well, because RCS' encryption is MLS. Security folks correct me if I'm wrong
Apple will soon support encrypted RCS messaging with Android users
Building bridges without blue bubbles.
www.theverge.com
March 15, 2025 at 12:43 AM
So one good thing that seems to be happening right now is that a new end-to-end encryption standard "MLS" seems to be gaining a lot of momentum. Like, a lot.
And from what I understand this is an important step there as well, because RCS' encryption is MLS. Security folks correct me if I'm wrong
And from what I understand this is an important step there as well, because RCS' encryption is MLS. Security folks correct me if I'm wrong
Reposted by Walter Hernandez
Wow, this seems to be extremely easy to code and extremely useful.
Transformers without Normalization
Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu
arxiv.org/abs/2503.10622
Transformers without Normalization
Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu
arxiv.org/abs/2503.10622
March 14, 2025 at 5:42 PM
Wow, this seems to be extremely easy to code and extremely useful.
Transformers without Normalization
Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu
arxiv.org/abs/2503.10622
Transformers without Normalization
Jiachen Zhu, Xinlei Chen, Kaiming He, Yann LeCun, Zhuang Liu
arxiv.org/abs/2503.10622
Reposted by Walter Hernandez
NEW 🧵 Is human intelligence starting to decline?
Recent results from major international tests show that the average person’s capacity to process information, use reasoning and solve novel problems has been falling since around the mid 2010s
What should we make of this?
www.ft.com/content/a801...
Recent results from major international tests show that the average person’s capacity to process information, use reasoning and solve novel problems has been falling since around the mid 2010s
What should we make of this?
www.ft.com/content/a801...
March 14, 2025 at 1:18 PM
NEW 🧵 Is human intelligence starting to decline?
Recent results from major international tests show that the average person’s capacity to process information, use reasoning and solve novel problems has been falling since around the mid 2010s
What should we make of this?
www.ft.com/content/a801...
Recent results from major international tests show that the average person’s capacity to process information, use reasoning and solve novel problems has been falling since around the mid 2010s
What should we make of this?
www.ft.com/content/a801...
Reposted by Walter Hernandez
We'll commit to a slice 🥧
Happy Pi Day!
Happy Pi Day!
March 14, 2025 at 5:27 PM
We'll commit to a slice 🥧
Happy Pi Day!
Happy Pi Day!
Reposted by Walter Hernandez
"Junk papers proliferate at vanity journals and legitimate ones alike, due in part to the “publish or perish” ethos that pervades the research enterprise, and in part to the catastrophic business model that has captured much of scientific publishing since the early 2000s."
"The scientific literature is an essential ocean of knowledge, in which floats an alarming amount of junk."
Reflecting on RFK Jr.'s use of scholarly papers in his confirmation hearings.
Reflecting on RFK Jr.'s use of scholarly papers in his confirmation hearings.
The Scientific Literature Can’t Save Us Now
You can cite peer-reviewed research in support of almost any claim, no matter how absurd.
www.theatlantic.com
February 15, 2025 at 9:44 AM
"Junk papers proliferate at vanity journals and legitimate ones alike, due in part to the “publish or perish” ethos that pervades the research enterprise, and in part to the catastrophic business model that has captured much of scientific publishing since the early 2000s."
Reposted by Walter Hernandez
It is so strange that we have to figure out how (or even whether) our latest software does critical functions that would normally have to be carefully designed.
More like biology or psychology than computer science.
More like biology or psychology than computer science.
March 8, 2025 at 7:28 PM
It is so strange that we have to figure out how (or even whether) our latest software does critical functions that would normally have to be carefully designed.
More like biology or psychology than computer science.
More like biology or psychology than computer science.
Reposted by Walter Hernandez
“We should stop training scientists now. It’s obvious that within three years, AI is going to do better than Nobel Laureates.”
is the new
“We should stop training radiologists now. It’s just completely obvious that within five years, deep learning is going to do better than radiologists.”
is the new
“We should stop training radiologists now. It’s just completely obvious that within five years, deep learning is going to do better than radiologists.”
March 8, 2025 at 6:31 PM
“We should stop training scientists now. It’s obvious that within three years, AI is going to do better than Nobel Laureates.”
is the new
“We should stop training radiologists now. It’s just completely obvious that within five years, deep learning is going to do better than radiologists.”
is the new
“We should stop training radiologists now. It’s just completely obvious that within five years, deep learning is going to do better than radiologists.”
Reposted by Walter Hernandez
Mistral Small 3
A 24B LLM that's VERY fast with great function calling
More important, MISTRAL IS OPEN SOURCE AGAIN!!!!!!
mistral.ai/news/mistral...
A 24B LLM that's VERY fast with great function calling
More important, MISTRAL IS OPEN SOURCE AGAIN!!!!!!
mistral.ai/news/mistral...
January 30, 2025 at 9:53 PM
Mistral Small 3
A 24B LLM that's VERY fast with great function calling
More important, MISTRAL IS OPEN SOURCE AGAIN!!!!!!
mistral.ai/news/mistral...
A 24B LLM that's VERY fast with great function calling
More important, MISTRAL IS OPEN SOURCE AGAIN!!!!!!
mistral.ai/news/mistral...
Reposted by Walter Hernandez
Interesting paper that tests GPT-4o’s ability to handle financial predictions and finds weak numeric reasoning & that a lot of apparent ability is actually due to memorized training data. At the same time, they show promise when combined with tool use. papers.ssrn.com/sol3/papers....
January 29, 2025 at 10:58 PM
Interesting paper that tests GPT-4o’s ability to handle financial predictions and finds weak numeric reasoning & that a lot of apparent ability is actually due to memorized training data. At the same time, they show promise when combined with tool use. papers.ssrn.com/sol3/papers....