How do chat models distill thousands of tweets into a coherent latent persona?
teortaxesTex · x · 2026-11-17
teortaxesTex raises a mechanistic puzzle about language models: when he asks the chat product Fable who he is, it produces a reasonably accurate portrait of his account. He wonders at what stage the mass of tweets coalesces into a cohesive latent persona.
The quoted tweet merely confirms he is an internet personality, adding no substance.
More from Models
- Devs: Gemini 4 Carbon is frontier for ML engineering, not deliberately crippled — olcan · 2026-11-17
- Ripple-65B builds its own local IDE in 3 hours at hackathon, >300 tok/s codegen — Azaliamirh · 2026-11-17
- OpenAI kicks off first phase of embedded assessments, next model may be weeks away — AP_in_Indy · 2026-11-17
- NVIDIA launches Nemotron for Science: fully open models, datasets and agent recipes — NVIDIAAI · 2026-11-17
- Diffusion images lack controllability: researchers argue for painter-style, tool-call-based generation — josephdviviano · 2026-11-17
- User Has Spent $2,513 on ChatGPT Since 2023 but Eyes a Switch to Claude — MoreLunch1547 · 2026-11-17