User repository on GitHub
Add as much tabular data as you want to Jev models using adaptive ensembles that learn to query only the rows and columns needed.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Repos & SDKs
Xjevboost
Classify tabular data with Jev, Laya or OpenJev without putting your whole labeled dataset into context: a learned tree of small views asks for more rows and columns only when they help, then stops.
- Stars
- 1
- Language
- Python
- License
- APACHE-2.0
- Reading time
- 2 min
- Published by
- Community
- Author
- sedthh
- Last verified
- Oct 2026
Works forJev for Classification
Used inJev for Data & Analytics
Highlights
- Does not retrain the model: a greedy tree search learns which views of your table to send and when to stop asking for more.
- Each requested view is one model call or a cached answer; a row may need one call or several within your call and token budgets.
- Same classifier works with Jev, Laya and OpenJev through swappable providers; Laya and OpenJev never borrow your TypeSafe key.
- Configured with n_views, max_depth, calibration_fraction, pruning_fraction, max_training_calls and random_state.
- Despite the name, no gradient boosting yet; the README says it is likely coming in a future version.
Quickstart
bash
pip install -e .
export TYPESAFE_API_KEY="your-api-key"Watch out
Grounded in the truncated README and GitHub metadata reviewed on 2026-10-15; the README cuts off in its final section, publishes no benchmark numbers, describes the goal as approaching full-context quality while spending fewer tokens, and warns token estimates stay approximate without a tokenizer-based bound.
Questions about this entry
Short answers drawn from the linked project or page.
It classifies tabular data with Jev, Laya or compatible OpenJev servers without putting your entire labeled dataset into context. Instead it learns which small slices (views) of the table to send, asking for more only when it helps.
A view contains selected labeled examples and columns plus the row to classify. A greedy tree search learns which views help and when to stop; each model answer guides whether to stop or request another view, within your call and token budgets.
Jev through JevProvider, Laya through LayaProvider and OpenJev through OpenJevProvider. You run the model server separately; use a model version and context allowance available to your account, and set context_limit to the deployed checkpoint's allowance.
No. The README is explicit: the model you provide is not retrained; xjevboost learns which context to send it. Despite the name, there is no gradient boosting yet, though the README says it is likely coming in a future version.
Each requested view is one model call, or a cached answer is reused. A row may need one call or several; routing tests use answers already available and make no extra calls. The README publishes no token or price numbers.
It identifies the server version, weights and inference configuration for caching. It does not pin weights on the server; change it when the deployment changes, and refit the policy when you switch to a different model.
Token estimates stay approximate unless you supply a tokenizer-based bound, and encoder models with short contexts need small views because a server may truncate oversized inputs. The digest's README is cut off before the end, so some sections were not reviewed.
More Jev resources like this
A Python batching proxy for Jev that returns Apache Arrow streams, with Python and JavaScript SDKs. It packs many independent states and questions into each TypeSafe API call or runs them concurrently, then emits the answers as an ordered Arrow IPC stream with the schema first.
Jev Repos & SDKs#community#github#python
Community
columnar-tech/jevaroPython
columnar-tech/jevaroPythonPython with a smart if: .jev files add the ~ question operator and jev/case blocks that Jev answers with a number or a label, never text. Files are rewritten to plain Python at import time; every ~ is one paid TypeSafe API call.
Jev Repos & SDKs#community#catalog-import#logicrw
Community
sumanmichael/jevlangPython
sumanmichael/jevlangPythonA Python and JavaScript package that loads any Hugging Face dataset, auto-adapts evaluation rubrics to the domain, and classifies every row through a single parallel Jev call with calibrated probabilities, plus a CLI and an interactive studio for uploads and export.
Jev Repos & SDKs#community#catalog-import#logicrw
Community
hemanth/hfjevTypeScript
hemanth/hfjevTypeScriptA System One layout optimizer for decision-model workloads: it reorders rows and fields so a prefix-caching backend reuses input computation, then returns per-record decisions in original order. Its bundled backend serves AutoTrust JEV-9B through vLLM.
Jev Repos & SDKs#community#github#python
Community
0814wdwd/solo_jevPython
0814wdwd/solo_jevPythonA Python library that turns Jev-style decision models into ordinary function calls: check returns a bool, classify picks one option, label picks every option that applies, and score rates on a scale, with probabilities behind every answer.
Jev Repos & SDKs#community#github#python
Community
jlowin/vibecheckPython
jlowin/vibecheckPythonA Python library that runs LLM and Jev calls as Polars expressions: parallel inference across providers, Pydantic-validated outputs as struct columns, and a native TypeSafe System One layer for per-row decision questions.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
pnthn-ai/polar_llamaPython
pnthn-ai/polar_llamaPythonFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.