Reading notes for grown-ups
Short educational notes for adults. Titles and identifiers only.
Educational and informational. Citations are limited to titles, authors, identifiers, and the principle phrases named below. This page does not restate figures or results from the sources.
For adults. These are reading notes, not activities for children, and not clinical advice. Do not enter a child’s name or other information about a child.
A founder guide to raw sessions, pointers, and receipts
Source: a publicly circulated founder-guide infographic about raw sessions, pointers, and receipts. This page records only the principle phrases below. Citation of that guide is not an affiliation or endorsement.
- Store raw sessions untouched.
- Keep pointers and receipts.
- The agent does not get to invent truth.
- Send code to data.
Test-Time Compute Scales Better with Parallel Agent Teams
Source: Junlin Wang, Haozhe Chen, Ang Li, Rahul Chalamala, and Zhuang Liu. “Test-Time Compute Scales Better with Parallel Agent Teams.” arXiv:2611.09798v1.
- The titled claim is that test-time compute scales better with parallel agent teams.
- A principle phrase recorded here is “small parallel teams plus a verifier.”
How to Train a Swarm
Source: General Reasoning, “How to Train a Swarm,” 2 November 2026. https://gr.inc/blog/how-to-train-a-swarm.
- Latency pressure.
- Communication primitives.
The Lottery Ticket Hypothesis
Source: Frankle and Carbin. ICLR 2019. arXiv:1803.03635v5. Only the public title, authors, venue, and identifier are used here. No results or numbers are added.
- This page records the public title and identifier only.
- This page does not restate the paper’s results, accuracy figures, or pruning percents.
LittleBit: Ultra Low-Bit Quantization via Latent Factorization
Source: Banseok Lee, Dongkyu Kim, Youngcheon You, and Youngmin Kim, Samsung Research. “LittleBit: Ultra Low-Bit Quantization via Latent Factorization.” arXiv:2506.13771. Citation of the authors and affiliation is not an endorsement or affiliation with Samsung Research.
- The titled subject is ultra low-bit quantization via latent factorization.
- No bit-width or benchmark numbers are stated on this page.
Pointer list for these notes: /listings/.
Last revised: 11 November 2026 (America/New_York). Educational page. Education and information only. For adults. This site is not directed at children under 13 and collects no information from children. No professional advice. Results are not guaranteed.