RememberboxA text-first reading library | Saved | 751 |
|---|---|---|
| Page | 29 |
2026-01-23
| Article | Read |
|---|---|
| Kona: Energy-Based Models (EBMs) for AI Reasoning Kona delivers AI reasoning via Energy-Based Models (EBMs). It provides deterministic, verifiable intelligence for critical systems—a fundamental shift from probabilistic LLMs. | |
| github.com |
2026-01-21
| Article | Read |
|---|---|
| The Mind of Napoleon by Christopher Herold This is not a source book but a selection of written and spoken utterances on a variety of subjects, grouped according to broad themes and designed to give the reader an insight into the mind of a man who combined energy of thought and energy of action to an exceptional degree. It showcases his almost supernatural power of mind. it demonstrates his ability to reduce all problems to their simplest elements and discard all obstacles to action – his amazing skill at learning and conquering any subj | |
| www.kickstarter.com |
2026-01-20
| Article | Read |
|---|---|
| Security Verification Security verification page to protect against malicious bots. | |
| docs.learnbittensor.org |
2026-01-18
| Article | Read |
|---|---|
| The 7 levels of AI – CASTLE Warren B. PowellProfessor Emeritus, Princeton University | |
| docs.nvidia.com | |
| developer.download.nvidia.com | |
| Work and Depth Analyzing performance is a key part of studying algorithms. Although such analysis is not used to predict the exact running time of an algorithm on a particular machine, it is important in determining how the running time grows as a function of the input size. To analyze performance a formal model is needed to account for the costs. In parallel computing the most common models are based on a set of processors connected either by a shared memory, as in the Parallel Random Access Machines (PRAM, s | |
| ghostty-shaders/bettercrt.glsl at main · 0xhckr/ghostty-shaders A repository containing many free shaders to use with ghostty (the terminal) - 0xhckr/ghostty-shaders | |
| Fun with Ghostty Shaders Back when Ghostty released I played around with the entire config, including trying to get some shaders to work. iTerm2 has the ability to have an image background in your terminal and ghostty does not, at least not directly. I wanted to get a custom image with a shader but couldn’t get it working. |
2026-01-16
| Article | Read |
|---|---|
| GitHub - vercel-labs/json-render: The Generative UI framework The Generative UI framework. Contribute to vercel-labs/json-render development by creating an account on GitHub. | |
| GitHub - t4sk/notes Contribute to t4sk/notes development by creating an account on GitHub. |
2026-01-14
| Article | Read |
|---|---|
| Application binary interface From Wikipedia, the free encyclopedia |
2026-01-13
| Article | Read |
|---|---|
| Triton Linear Layout: Concept | Lei.Chat() <p>Layout is a core concept in Triton for representing and optimizing distribution mappings from source problems to the target hardware compute and memory hierarchy. In this blog post I will talk about linear layout in Triton, the new unifying mechanism over existing bespoke layouts for different purposes. The aim is to provide motivation and an intuitive understanding of linear layout; I will rely on examples and illustrations instead of theories and proofs.</p> | |
| Checking your browser - reCAPTCHA Checking your browser before accessing pmc.ncbi.nlm.nih.gov ... | |
| Ganoderma lucidum polysaccharides ameliorated depression-like behaviors in the chronic social defeat stress depression model via modulation of Dectin-1 and the innate immune system - PubMed Major depressive disorder (MDD) is a prevalent, chronic, and recurrent disease. At least one-third of patients have treatment-resistant depression; therefore, there is an urgent need for novel drug development. Cumulative studies have suggested an inflammatory mechanism for the pathophysiology of MD … |
2026-01-12
| Article | Read |
|---|---|
| Checking your browser - reCAPTCHA Checking your browser before accessing pmc.ncbi.nlm.nih.gov ... |
2026-01-11
| Article | Read |
|---|---|
| Breaking the Sorting Barrier for Directed Single-Source Shortest Paths We give a deterministic $O(m\log^{2/3}n)$-time algorithm for single-source shortest paths (SSSP) on directed graphs with real non-negative edge weights in the comparison-addition model. This is the first result to break the $O(m+n\log n)$ time bound of Dijkstra's algorithm on sparse graphs, showing that Dijkstra's algorithm is not optimal for SSSP. | |
| www.k-a.in |
2026-01-10
| Article | Read |
|---|---|
| GitHub - parcadei/Continuous-Claude-v3: Context management for Claude Code. Hooks maintain state via ledgers and handoffs. MCP execution without context pollution. Agent orchestration with isolated context windows. Context management for Claude Code. Hooks maintain state via ledgers and handoffs. MCP execution without context pollution. Agent orchestration with isolated context windows. - parcadei/Continuous-... | |
| Deep Delta Learning Transformer residual streams evolve by additive accumulation: each layer appends a feature update to a shared hidden state, but has no direct mechanism for replacing content that has become obsolete or conflicting. We introduce Deep Delta Learning (DDL), a residual update rule that preserves the identity path while giving every layer the ability to selectively rewrite residual content. DDL reads the current state along a learned direction, compares it with a learned target value, and writes back | |
| GitHub - yifanzhang-pro/deep-delta-learning: Official Project Page for Deep Delta Learning (https://arxiv.org/abs/2601.00417) Official Project Page for Deep Delta Learning (https://arxiv.org/abs/2601.00417) - yifanzhang-pro/deep-delta-learning |
2026-01-09
| Article | Read |
|---|---|
| Emergent temporal abstractions in autoregressive models enable hierarchical reinforcement learning Large-scale autoregressive models pretrained on next-token prediction and finetuned with reinforcement learning (RL) have achieved unprecedented success on many problem domains. During RL, these models explore by generating new outputs, one token at a time. However, sampling actions token-by-token can result in highly inefficient learning, particularly when rewards are sparse. Here, we show that it is possible to overcome this problem by acting and exploring within the internal representations o |