User repository on GitHub
I tortured Jev into being a RISC-V CPU.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Practices & Patterns
RISC-jeV
A stunt build where C is compiled and run on a simulated SERV RISC-V CPU whose logic gates are lookup tables of 22 boolean answers that Jev fills in, with the clock, trace and cost on screen.
- Stars
- 1
- Language
- Python
- Published by
- Community
- Author
- i2cjak
- Last verified
- Oct 2026
About this project
Jev is Turing complete. I made Jev parse boolean expressions. I tortured Jev into being a RISC-V CPU. Write C, compile it, and run it on a simulated SERV CPU. Watch the clock, instruction trace, signals, Jev requests, and cost. Jev answers 22 Boolean gate/input combinations. Those answers become cached lookup tables used by every simulated logic gate. Cached runs make no new API calls. Fresh Jev replaces the tables with a new response. No correctness checks or confidence threshold.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- You write C, it runs on a simulated SERV CPU, and the interface shows the clock, instruction trace, signals, Jev requests and cost.
- Jev answers 22 boolean gate and input combinations once, and every simulated logic gate then reads from those cached tables.
- Cached runs make no new API calls; asking for a fresh Jev run replaces the tables with a new set of answers.
- Each run is capped at five minutes and an estimated $0.05 input budget at $0.042 per million tokens.
Watch out
This is a demonstration, not a tool: the README frames it as making Jev parse boolean expressions and turning it into a RISC-V CPU, and there are no correctness checks or confidence threshold anywhere in the loop, so nothing detects a bad set of 22 answers. Cached runs reuse whatever was fetched earlier, which means results are only as sound as those answers. The budget cap covers estimated input only, output charges are not included, and the run stops before sending a batch that might exceed the remaining budget. The README is about twenty lines long and defers local setup to docs/architecture.md, so expect to read that before running it yourself. 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.
A demo, and an entertaining one. The README's own framing is that Jev is Turing complete, that the author made it parse boolean expressions, and that he tortured it into being a RISC-V CPU. The interesting claim is compositional: a handful of bounded yes/no questions about gate inputs is enough to drive a simulated machine. It is not something to build a toolchain on.
Twenty-two boolean gate and input combinations, answered once. Those answers become cached lookup tables that every simulated logic gate consults, so a cached run makes no new API calls at all. Requesting a fresh Jev run replaces the tables with a new response, which is the moment the cost and latency appear.
Each run gets up to five minutes and an estimated $0.05 input budget at $0.042 per million tokens. The runner stops before sending a batch that may exceed the remaining budget, so you get a partial run rather than an overrun. Output charges are not included in that estimate, which the README calls out explicitly.
Nothing detects it. The README is plain that there are no correctness checks and no confidence threshold, so a wrong boolean propagates into the lookup tables and from there into the simulated logic. There is no verification step and no way to tell a correct circuit from a plausible one without checking the trace yourself.
You write C, it is compiled, and it runs on a simulated SERV CPU. The interface shows the clock, the instruction trace, the signals, the Jev requests and the running cost, so the effect of each new set of 22 answers is visible as the program steps. There is also a hosted instance at jev-riscv-production.up.railway.app if you would rather not build anything.
The README is about twenty lines and defers all local setup to docs/architecture.md. There is no documented package-manager step, no key configuration and no build command in the README itself, so expect to read that architecture document before running it. The hosted link is the low-friction path, and it is what the project points you at first.
SERV, the RISC-V core linked from olofk, and it is a simulator rather than real hardware. Your C compiles for that target, and every logic gate inside it reads from the lookup tables Jev filled in, so what you observe is the simulated core acting on 22 boolean answers rather than a physical chip running your program.
More Jev resources like this
24GitHub stars
Unofficial laptop study of parallel constrained decoding: prefill once, share the KV cache across fields, and decide all schema fields in one batched forward pass without writing JSON.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
rorshopping/jev-on-a-laptopPython
rorshopping/jev-on-a-laptopPython22GitHub stars
Offline evaluator that checks whether classifier confidences match reality and converts mistake costs into a deployable threshold, emitting an HTML report plus machine-readable threshold files.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
rlaope/jevalPython
rlaope/jevalPython16GitHub stars
Vendor-neutral harness that tests whether Jev reranking improves a small RAG stack, using frozen hybrid BM25 plus BGE-M3 candidates and isolating retrieval, rerank, and generation stages.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
erendikmenn/jev-rag-benchmarkPython
erendikmenn/jev-rag-benchmarkPython37GitHub stars
Minimal Python recipe that turns any causal LLM into a Jev-style classifier by prefilling choice_index and constraining sampling to legal indexes in one forward pass.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
sgoedecke/system-onePython
sgoedecke/system-onePython39GitHub stars
Open replica of the typed decision API on openly available base models, answering choice, score, and true-probability questions in one forward pass with no decoding.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
mithalouni/system-one-openPython
mithalouni/system-one-openPython34GitHub stars
Curated catalog of community Jev applications grouped by use, where each case explains the mechanism and states what the public evidence does and does not prove.
Jev Practices & Patterns#community#jevbest#catalog-import
Community
JackZeng/Jev_appsPython
JackZeng/Jev_appsPythonFrom the Jev guides
Original write-ups on what this entry is for.
Most Jev accuracy problems are state problems. What belongs in a state, what is a pre-made judgment to leave out, and why naming the slice you ask about matters more than the wording.
Jev Practices & Patterns#state#design#primitives
Read article
A Noul returns one number and no separate confidence. How to read the value, pin its boundary with criteria, split compound questions, and set a per-action threshold that matches the cost of being wrong.
Jev Practices & Patterns#noul#primitives#confidence
Read article
A Score is a weighted position, not a ruler. Why the number can land between levels, why levels are judged separately, and how to combine several atomic scores in code instead of one crowded rubric.
Jev Practices & Patterns#score#primitives#scoring
Read article
From the Jev community
Posts from builders shipping with Jev right now.