Post
140
Much of the misunderstanding on loops and byte-level models comes from fear on inventing the wheel, not reinventing it
ZeroGPU Explorers
community
AI & ML interests
None defined yet.
Recent Activity
Post
135
How much should OpenAI offer for a name like Dot Labs?
- 1 reply
soujanyaporia
authored 4
papers 4 days ago
BaRe-Mem: Bayesian Reliability Memory for Robust and Adaptive Agent Consultation
Paper • 2609.35551 • Published • 8
Just MLPs: Efficient Visual State Reconstruction for Multimodal Language Models
Paper • 2609.34972 • Published • 30
MemBodied: Recurrent Associative Memory for Vision-Language-Action Models
Paper • 2609.28256 • Published • 15
Chain-of-Knowledge: Grounding Large Language Models via Dynamic Knowledge Adapting over Heterogeneous Sources
Paper • 2305.13269 • Published
Post
165
Reality has not limits if you know how to shape it, one step at a time.
- 1 reply
Post
84
How old were you when you discovered good AI agents favor the Strategy Design Pattern over any other?
- 1 reply
soujanyaporia
submitted a
paper to Daily Papers 9 days ago
BestWishYsh
authored a
paper 11 days ago
Post
131
LLMS for edge devices?
What if we go the reliable route instead of the speed route?
What if we make it run on sbcs with few megabytes available?
That's the idea for the next model.
Keep in tune.
What if we go the reliable route instead of the speed route?
What if we make it run on sbcs with few megabytes available?
That's the idea for the next model.
Keep in tune.
Post
141
We trained a 10.9M byte-level recurrent Transformer on L3 and L6. (Loop 3 and Loop 6)
Yet L4/L5 improved too, L8 held up, and the L3→L6 gain grew during training.
Same weights. More compute. Better predictions.
This is a new architecture for effective compute after several steps beyond original training!
We mixed and matched components like time and mhc into an ouro-like byte-level language model and the result is BET, a byte-level step-elastic transformer that can run computation steps without significant degradation.
One of the coolest parts of this training was discovering how Gradient Descent decided to use the first layer as what we would consider a scratchpad! Totally destroyed for the decoder but somehow makes total sense for the next layer!
I believe looped-transformers are the future of edge computing and this is a first step towards it.
Blogpost: https://medium.com/@appvoidofficial/byte-level-elasticity-182fe2ed1d2f
appvoid/bet-10m
Yet L4/L5 improved too, L8 held up, and the L3→L6 gain grew during training.
Same weights. More compute. Better predictions.
This is a new architecture for effective compute after several steps beyond original training!
We mixed and matched components like time and mhc into an ouro-like byte-level language model and the result is BET, a byte-level step-elastic transformer that can run computation steps without significant degradation.
One of the coolest parts of this training was discovering how Gradient Descent decided to use the first layer as what we would consider a scratchpad! Totally destroyed for the decoder but somehow makes total sense for the next layer!
I believe looped-transformers are the future of edge computing and this is a first step towards it.
Blogpost: https://medium.com/@appvoidofficial/byte-level-elasticity-182fe2ed1d2f
appvoid/bet-10m
soujanyaporia
authored 8
papers 23 days ago
GE-Act 2.0: Pretraining and Scaling a World-Action Model for Robotic Manipulation
Paper • 2609.05588 • Published • 57
MNIST-PRO: MNIST is Back as a Partially Observable World for AI Agents
Paper • 2608.31022 • Published • 9
ScrambleToolBench: Agents Search Exhaustively Even When Their Own Map Points to the Next Step
Paper • 2608.02358 • Published • 12
Data Agent: Learning to Select Data via End-to-End Dynamic Optimization
Paper • 2603.07433 • Published
$Σ$-Mem: An Online Reliability Memory for LLM-based Multi-Agent Systems
Paper • 2607.27958 • Published • 19
IDEAgent: Agentic Quality-Diversity Search for Research Idea Generation
Paper • 2607.22375 • Published • 10
On the Limits of LLM-as-Judge for Scientific Novelty Assessment
Paper • 2606.12071 • Published • 2
GRAIL: Gradient-Reweighted Advantages for Reinforcement Learning with Verifiable Rewards
Paper • 2606.04889 • Published • 4