User repository on GitHub
Selenium 2048 player powered by expectimax search and TypeSafe Jev, with portrait FFmpeg recording.
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Demos & Experiments
Jev 2048 Selenium
Selenium player for 2048.org that reads the exact board state, runs time-bounded expectimax over the 90% 2 / 10% 4 spawn distribution, and fuses it with one Jev Choice per move.
- Stars
- 3
- Language
- Python
- Published by
- Community
- Author
- AMMIROSOH
- Last verified
- Oct 2026
About this project
This Python application opens 2048.org with Selenium and plays automatically. The default hybrid mode combines a time-bounded expectimax search with a real TypeSafe Jev Choice decision on every move. It also includes an integrated Selenium + FFmpeg portrait recorder. See RECORDING.md, or run python run.py --record to create a high-quality 1080×1920 YouTube Short MP4 automatically. Star The Repo if you Enjoyed ⭐. Jev 1.13 is a text/JSON decision model, not a vision or game-tree model. The application therefore:
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- Hybrid mode combines iterative-deepening expectimax with one bounded Jev Choice.
- Confidence-gated fusion caps Jev weight so uncertain answers cannot discard best moves.
- Selenium plus FFmpeg recorder produces portrait YouTube Short MP4 via --record.
- Per-move JSON Lines logs capture board, candidates, model version and failures.
Watch out
Random spawns mean no guaranteed tile or score; pin a documented model when benchmarking. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
Hybrid fuses search and Jev distributions with Jev confidence controlling its weight, so uncertain answers cannot override a clearly superior search move. Strict jev mode gives Jev final authority for experiments.
It reads the exact 4x4 board from saved game state with DOM fallback, reproduces merge and scoring locally, removes illegal moves, then sends only legal moves plus search evidence.
Python 3.10 plus Chrome, Edge or Firefox with Selenium Manager handling the driver, TYPESAFE_API_KEY in the environment, and python run.py. Offline smoke tests use --mode search with no key.
Every move logs board, search candidates, Jev choice and confidence, fused probabilities and failures as JSON Lines under logs/. Network errors fail open to search and stop calling Jev after three in a row.
hybrid (default) fuses a time-bounded expectimax search with a Jev Choice per move, weighted by Jev's confidence; jev gives Jev final authority over the legal move list; search runs offline expectimax with no API calls.
Python 3.10 or newer and Chrome, Edge, or Firefox; Selenium Manager downloads or locates the matching driver automatically. --browser edge and --search-ms tune the run.
python run.py --mode search --headless --max-moves 20 is an offline Selenium and search test that needs no API key. Every move is logged as JSON Lines under logs/ with the board, candidates, Jev output, and any API failure.
More Jev resources like this
46GitHub stars
Open 150M encoder that reproduces the typed noul, choice, and score interface in one forward pass, trainable on a free GPU and runnable locally for nothing.
Jev Demos & Experiments#community#jevbest#catalog-import
941GitHub stars
Library that adds schema-guaranteed typed outputs to existing autoregressive LLMs without changing weights, served through SGLang or a hosted API.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
TypeLLM/TypeLLMPython
TypeLLM/TypeLLMPython55GitHub stars
Toolkit that makes a model behind an OpenAI-compatible endpoint or running in-process answer typed choice, score, and yes/no questions from next-token probabilities in Jev API shape.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
fidecastro/jevifyPython
fidecastro/jevifyPython164GitHub stars
End-to-end toolkit that synthesizes decision data, trains and calibrates candidate scorers, evaluates fixed splits, and serves Jev-compatible inference for interactive web decisions.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
zwliJay/jev-forgePython
zwliJay/jev-forgePython63GitHub stars
Claude Code kit that guards PreToolUse calls with a free pre-filter plus one Jev question for ambiguous cases, and adds sub-agent sizing, rechecks, search, daemon speedups, and monitoring.
Jev Demos & Experiments#community#awesomejev#catalog-import
Community
jonathanavis96/jev-kitPython
jonathanavis96/jev-kitPython70GitHub stars
Self-hosted proxy that records typed decisions, distils them into small heads on a frozen encoder, and answers locally when confident.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
bladedevoff/stuntdPython
bladedevoff/stuntdPythonFrom the Jev guides
Original write-ups on what this entry is for.
A game tick is a state plus a legal action list. How Clash Royale, Minecraft, and the earlier Mario, Pokémon, StarCraft, and drone loops ask Jev one step at a time.
Jev Demos & Experiments#gaming#agents#control
Read article
From the Jev community
Posts from builders shipping with Jev right now.