I'm giving a GDC talk this year!
March 8, 2025 at 12:47 AM
I'm giving a GDC talk this year!
In @science.org
www.science.org/doi/10.1126/...
Human language & whale song share statistical structure for ‘learnability’
#culture #whale
@ellengarland.bsky.social @jennyallen13.bsky.social @inbalarnon.bsky.social @simonkirby.bsky.social @clairenea.bsky.social @aucklanduni.bsky.social
www.science.org/doi/10.1126/...
Human language & whale song share statistical structure for ‘learnability’
#culture #whale
@ellengarland.bsky.social @jennyallen13.bsky.social @inbalarnon.bsky.social @simonkirby.bsky.social @clairenea.bsky.social @aucklanduni.bsky.social
February 6, 2025 at 7:28 PM
In @science.org
www.science.org/doi/10.1126/...
Human language & whale song share statistical structure for ‘learnability’
#culture #whale
@ellengarland.bsky.social @jennyallen13.bsky.social @inbalarnon.bsky.social @simonkirby.bsky.social @clairenea.bsky.social @aucklanduni.bsky.social
www.science.org/doi/10.1126/...
Human language & whale song share statistical structure for ‘learnability’
#culture #whale
@ellengarland.bsky.social @jennyallen13.bsky.social @inbalarnon.bsky.social @simonkirby.bsky.social @clairenea.bsky.social @aucklanduni.bsky.social
Incorporating a scientifically strong conceptual hierarchy may increase learnability and applicability
#HierarchicalLearningFramework
#HierarchicalLearningFramework
The Editor-in-Chief of the journal Science thinks undergrad science majors should be taught philosophy of science. www.science.org/doi/10.1126/...
Here at the University of Cincinnati, the Philosophy Department developed a course called How Science Works a decade ago...
Here at the University of Cincinnati, the Philosophy Department developed a course called How Science Works a decade ago...
Teach philosophy of science
Much is being made about the erosion of public trust in science. Surveys show a modest decline in the United States from a very high level of trust, but that is seen for other institutions as well. Wh...
www.science.org
September 8, 2024 at 11:23 PM
Incorporating a scientifically strong conceptual hierarchy may increase learnability and applicability
#HierarchicalLearningFramework
#HierarchicalLearningFramework
this is extremely deep in the LLM weeds, but I think that ALiBi and other positional encoding schemes are extremely weirdly theoretically motivated. the problem with the original sine embeddings was distinguishability, not learnability!
these word embeddings are then entangled with positional embeddings, which are just a periodic function where every word is assigned to its own period of the function. "the peak of a sine wave" might be such a period, although the way this is done now is somewhat more complex than that.
October 14, 2025 at 6:25 PM
this is extremely deep in the LLM weeds, but I think that ALiBi and other positional encoding schemes are extremely weirdly theoretically motivated. the problem with the original sine embeddings was distinguishability, not learnability!
I'm researching board game teachability and learnability for a talk! Especially interested in innovative techniques. Does anyone know of case studies or other resources that might be useful? Thanks! 🎲✂️
September 5, 2024 at 4:17 AM
I'm researching board game teachability and learnability for a talk! Especially interested in innovative techniques. Does anyone know of case studies or other resources that might be useful? Thanks! 🎲✂️
One of the hardest part of developing any model is prepping a highly curated dataset, so they propose the Absolute Zero paradigm.
Their reasoner learns to both propose tasks that maximize learnability and improve reasoning by solving them, entirely through self-play—with no external data!
Their reasoner learns to both propose tasks that maximize learnability and improve reasoning by solving them, entirely through self-play—with no external data!
May 9, 2025 at 1:48 AM
One of the hardest part of developing any model is prepping a highly curated dataset, so they propose the Absolute Zero paradigm.
Their reasoner learns to both propose tasks that maximize learnability and improve reasoning by solving them, entirely through self-play—with no external data!
Their reasoner learns to both propose tasks that maximize learnability and improve reasoning by solving them, entirely through self-play—with no external data!
i'm working on player-sized enemies which turn toward the player and drift a little. some of them (all of them?) are going to shoot periodically. it's interesting thinking about how to establish learnability and visual communication with generated enemies.
November 27, 2025 at 4:54 AM
i'm working on player-sized enemies which turn toward the player and drift a little. some of them (all of them?) are going to shoot periodically. it's interesting thinking about how to establish learnability and visual communication with generated enemies.
Remember Chomsky's arguments about language innateness based on its putative un-learnability? Those arguments hinge on a pretty crazy theory of grammar. Here is Ted Gibson's summary of the original claims and a proposed solution in the form of dependency grammar:
tinyurl.com/yzjrpbb8
tinyurl.com/yzjrpbb8
April 20, 2026 at 5:50 PM
Remember Chomsky's arguments about language innateness based on its putative un-learnability? Those arguments hinge on a pretty crazy theory of grammar. Here is Ted Gibson's summary of the original claims and a proposed solution in the form of dependency grammar:
tinyurl.com/yzjrpbb8
tinyurl.com/yzjrpbb8
in general, i would observe that modern games have a lot of merits, like learnability and truly wonderful tactility and embodiment.
but some discarded interaction modes are amazing in terms of making the world feel raw and alive.
but some discarded interaction modes are amazing in terms of making the world feel raw and alive.
January 19, 2026 at 1:12 AM
in general, i would observe that modern games have a lot of merits, like learnability and truly wonderful tactility and embodiment.
but some discarded interaction modes are amazing in terms of making the world feel raw and alive.
but some discarded interaction modes are amazing in terms of making the world feel raw and alive.
Learnability is a specialized online tutoring platform dedicated to helping students with learning differences thrive academically and personally.
thelearnability.com
thelearnability.com
Learning Disabilities Tutor - Multisensory Approach | Learnability
Grow with our online learning disabilities tutor. Learnability offers multisensory learning strategies for children, adults, and students. Enroll today!
thelearnability.com
July 14, 2025 at 8:31 AM
Learnability is a specialized online tutoring platform dedicated to helping students with learning differences thrive academically and personally.
thelearnability.com
thelearnability.com
I would go further and say that learnability presupposes a specific question and associated hypothesis space. With art, there's no agreed-upon question or hypothesis space, and usually when people try to come up with one it ends up being about something else.
October 15, 2025 at 9:55 AM
I would go further and say that learnability presupposes a specific question and associated hypothesis space. With art, there's no agreed-upon question or hypothesis space, and usually when people try to come up with one it ends up being about something else.
The learnability of sign language morphology.
April 9, 2025 at 7:12 PM
The learnability of sign language morphology.
Humans and whales are evolutionarily distant, but united in having culture. When we use what we know about language learning to look at whale song, its hidden structure is revealed. These two systems have evolved culturally to maximise their own learnability - statistical structure is the result.
February 7, 2025 at 12:32 PM
Humans and whales are evolutionarily distant, but united in having culture. When we use what we know about language learning to look at whale song, its hidden structure is revealed. These two systems have evolved culturally to maximise their own learnability - statistical structure is the result.
How to characterize the learnability of local algorithms ?
The Merged Staircase Property (MSP) proposed by Abbe et al. (2022) is used to completely characterize the learnability of SGD-trained 2-layer neural networks (NN) in the regime where mean-field approximation holds for SGD.
The Merged Staircase Property (MSP) proposed by Abbe et al. (2022) is used to completely characterize the learnability of SGD-trained 2-layer neural networks (NN) in the regime where mean-field approximation holds for SGD.
November 27, 2024 at 3:07 PM
How to characterize the learnability of local algorithms ?
The Merged Staircase Property (MSP) proposed by Abbe et al. (2022) is used to completely characterize the learnability of SGD-trained 2-layer neural networks (NN) in the regime where mean-field approximation holds for SGD.
The Merged Staircase Property (MSP) proposed by Abbe et al. (2022) is used to completely characterize the learnability of SGD-trained 2-layer neural networks (NN) in the regime where mean-field approximation holds for SGD.
"A Unified Theory of Supervised Online Learnability," by Vinod Raman, Unique Subedi, and Ambuj Tewari. 2/5
arxiv.org/abs/2307.03816
arxiv.org/abs/2307.03816
A Combinatorial Characterization of Supervised Online Learnability
We study the online learnability of hypothesis classes with respect to arbitrary, but bounded loss functions. No characterization of online learnability is known at this level of generality. We give a...
arxiv.org
February 24, 2025 at 12:38 PM
"A Unified Theory of Supervised Online Learnability," by Vinod Raman, Unique Subedi, and Ambuj Tewari. 2/5
arxiv.org/abs/2307.03816
arxiv.org/abs/2307.03816
*Press A for Assistance: Making Games Accessible with AI* by Nonika & Steven Rice (U. Windsor) explores how AI can improve accessibility, approachability & learnability to make games more inclusive.
🎟️ Tickets: bit.ly/4nmKLzd
#AIandGames #GameDev #Accessibility #AI #Conference2025
🎟️ Tickets: bit.ly/4nmKLzd
#AIandGames #GameDev #Accessibility #AI #Conference2025
September 15, 2025 at 7:30 AM
*Press A for Assistance: Making Games Accessible with AI* by Nonika & Steven Rice (U. Windsor) explores how AI can improve accessibility, approachability & learnability to make games more inclusive.
🎟️ Tickets: bit.ly/4nmKLzd
#AIandGames #GameDev #Accessibility #AI #Conference2025
🎟️ Tickets: bit.ly/4nmKLzd
#AIandGames #GameDev #Accessibility #AI #Conference2025
The LIVE Programming Workshop will be held online Oct 17. Submissions due July 21. liveprog.org
Workshop on Live Programming (LIVE)
The 12th Workshop on Live Programming (LIVE 2026) will take place online. LIVE invites submissions of ideas for improving the immediacy, usability, and learnability of programming.
liveprog.org
February 6, 2026 at 9:43 PM
The LIVE Programming Workshop will be held online Oct 17. Submissions due July 21. liveprog.org
One more week to submit to LIVE, one of my favourite conferences!
Let's see your "new user interfaces that improve the immediacy, usability, and learnability of programming"
liveprog.org
Let's see your "new user interfaces that improve the immediacy, usability, and learnability of programming"
liveprog.org
One week until MONDAY, JULY 21st, which I call "LIVE Workshop Christmas".
(It's Christmas for me and the PC, cuz we get to look at all the cool stuff you submit! I guess for you it's more like "the submission deadline".)
liveprog.org
(It's Christmas for me and the PC, cuz we get to look at all the cool stuff you submit! I guess for you it's more like "the submission deadline".)
liveprog.org
July 15, 2025 at 6:59 AM
One more week to submit to LIVE, one of my favourite conferences!
Let's see your "new user interfaces that improve the immediacy, usability, and learnability of programming"
liveprog.org
Let's see your "new user interfaces that improve the immediacy, usability, and learnability of programming"
liveprog.org
Reminder: LIVE Programming workshop submissions due July 21. liveprog.org
Workshop on Live Programming (LIVE)
The 11th Workshop on Live Programming (LIVE 2025) will take place online. LIVE invites submissions of ideas for improving the immediacy, usability, and learnability of programming.
liveprog.org
June 16, 2025 at 1:44 PM
Reminder: LIVE Programming workshop submissions due July 21. liveprog.org
A study presents SOAR, a meta-RL framework empowering LLMs to create their own curricula, boosting performance on tough problems. This unique approach uses grounded rewards to kickstart learning, enabling reasoning capabilities to surpass traditional limits. https://arxiv.org/abs/2601.18778
Teaching Models to Teach Themselves: Reasoning at the Edge of Learnability
ArXiv link for Teaching Models to Teach Themselves: Reasoning at the Edge of Learnability
arxiv.org
January 28, 2026 at 2:31 AM
A study presents SOAR, a meta-RL framework empowering LLMs to create their own curricula, boosting performance on tough problems. This unique approach uses grounded rewards to kickstart learning, enabling reasoning capabilities to surpass traditional limits. https://arxiv.org/abs/2601.18778
just wanna commune w/my code or talk about it really don't wanna do anything else if anyone wanna talk about it a lot has happened recently!!!
readme not up to date super but is accurate *enough*
readme not up to date super but is accurate *enough*
GitHub - heartpunk/learnability
Contribute to heartpunk/learnability development by creating an account on GitHub.
github.com
March 12, 2026 at 10:41 PM
just wanna commune w/my code or talk about it really don't wanna do anything else if anyone wanna talk about it a lot has happened recently!!!
readme not up to date super but is accurate *enough*
readme not up to date super but is accurate *enough*
Ever wondered why "Where did you say that I was" sounds better than "??Where did you reiterate that I was"?
In this paper I co-authored with Julie Legate and Charles Yang, we present a learning-based account of this lexically-conditioned "island" effect (and the lack thereof in Mandarin)!
In this paper I co-authored with Julie Legate and Charles Yang, we present a learning-based account of this lexically-conditioned "island" effect (and the lack thereof in Mandarin)!
The learnability of bridge effects | Journal of Linguistics | Cambridge Core
The learnability of bridge effects
www.cambridge.org
July 5, 2026 at 12:26 AM
Ever wondered why "Where did you say that I was" sounds better than "??Where did you reiterate that I was"?
In this paper I co-authored with Julie Legate and Charles Yang, we present a learning-based account of this lexically-conditioned "island" effect (and the lack thereof in Mandarin)!
In this paper I co-authored with Julie Legate and Charles Yang, we present a learning-based account of this lexically-conditioned "island" effect (and the lack thereof in Mandarin)!
Readability ≠ Learnability by Ivan Lee & Taylor Berg-Kirkpatrick:
Developmentally plausible LM training works not because of simpler language but because of lower n-gram diversity! Warning against anthropomorphizing / equating learning in LMs and in children.
openreview.net/pdf?id=AFMGb...
Developmentally plausible LM training works not because of simpler language but because of lower n-gram diversity! Warning against anthropomorphizing / equating learning in LMs and in children.
openreview.net/pdf?id=AFMGb...
October 14, 2025 at 12:43 AM
Readability ≠ Learnability by Ivan Lee & Taylor Berg-Kirkpatrick:
Developmentally plausible LM training works not because of simpler language but because of lower n-gram diversity! Warning against anthropomorphizing / equating learning in LMs and in children.
openreview.net/pdf?id=AFMGb...
Developmentally plausible LM training works not because of simpler language but because of lower n-gram diversity! Warning against anthropomorphizing / equating learning in LMs and in children.
openreview.net/pdf?id=AFMGb...
The LIVE Programming Workshop will be held online this year. So no excuses - submit by July 21. liveprog.org
Workshop on Live Programming (LIVE)
The 11th Workshop on Live Programming (LIVE 2025) will take place online. LIVE invites submissions of ideas for improving the immediacy, usability, and learnability of programming.
liveprog.org
March 11, 2025 at 12:46 PM
The LIVE Programming Workshop will be held online this year. So no excuses - submit by July 21. liveprog.org
I’m pretty excited about this one!
ALTA is A Language for Transformer Analysis.
Because ALTA programs can be compiled to transformer weights, it provides constructive proofs of transformer expressivity. It also offers new analytic tools for *learnability*.
arxiv.org/abs/2410.18077
ALTA is A Language for Transformer Analysis.
Because ALTA programs can be compiled to transformer weights, it provides constructive proofs of transformer expressivity. It also offers new analytic tools for *learnability*.
arxiv.org/abs/2410.18077
ALTA: Compiler-Based Analysis of Transformers
We propose a new programming language called ALTA and a compiler that can map ALTA programs to Transformer weights. ALTA is inspired by RASP, a language proposed by Weiss et al. (2021), and Tracr (Lin...
arxiv.org
October 24, 2024 at 3:31 AM
I’m pretty excited about this one!
ALTA is A Language for Transformer Analysis.
Because ALTA programs can be compiled to transformer weights, it provides constructive proofs of transformer expressivity. It also offers new analytic tools for *learnability*.
arxiv.org/abs/2410.18077
ALTA is A Language for Transformer Analysis.
Because ALTA programs can be compiled to transformer weights, it provides constructive proofs of transformer expressivity. It also offers new analytic tools for *learnability*.
arxiv.org/abs/2410.18077