Jev directory
CommunityJev Models & Reimplementations
Sagicc/nanoGentzen-Jev
Hugging Face dataset of Gentzen LJ proof traces for intuitionistic logic, with per-step sequents, rules, pivots, and target values. Rows also carry tokenized input_ids and token_length, giving step-level supervision for decision models.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Hosted on Hugging Face as Sagicc/nanoGentzen-Jev and released under the MIT license.
- Listed tasks are text classification, reinforcement learning, and feature extraction.
- Rows carry sequent, rule, rule_idx, pivot, target_value, root_sequent, trace_step, and total_trace_steps.
- Example traces use the rules L_IMP and AXIOM with a total_trace_steps value of 5.
- Listed size is 1K-10K rows in JSON, English, with tabular and text modalities.
Questions about this entry
Short answers drawn from the linked project or page.
It is a Hugging Face dataset of Gentzen LJ proof traces for intuitionistic logic and automated reasoning. Each row records one step of a proof search, including the sequent, the rule applied, the pivot, and the target value, plus tokenized input_ids and a token_length.
Rows include sample_id, curriculum, sequent, rule, rule_idx, pivot, target_value, root_sequent, trace_step, total_trace_steps, input_ids, token_length, and isConversational. The curriculum field appears with values such as horn_policy_provable.
The dataset card lists an MIT license. That permits reuse and redistribution under the terms of that license, but you should confirm the current license text on the dataset page before relying on it for a shipped product.
The dataset is stored as JSON and is listed at a size of 1K to 10K rows. Its modalities are tabular and text, and the listed language is English.
The dataset card lists the Datasets library along with Dask and Croissant. Those are the loading and metadata surfaces named on the page, so Hugging Face Datasets is the most direct path to pull the rows.
The traces are tagged intuitionistic-logic and gentzen-lj, with additional tags for logic, automated-reasoning, theorem-proving, and decision-model. Example rows show the rules L_IMP and AXIOM applied to sequents built from implication clauses.
It is tagged decision-model and supplies per-step target_value labels over sequents, which is the step-level supervision shape a typed decision model consumes. It is a dataset, not a Jev model or SDK, so it is training material rather than a runtime component.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
Hugging Face dataset from IntelligenceLab holding 100K-1M GUI-agent decision samples for Jev-style models. Each row pairs screenshots with a goal state, a multiple-choice question, an answer index, and metadata such as action types, horizon, and negative-plan recipes.
Jev Models & Reimplementations#agents#open-models#classification
Community
Hugging Face dataset that converts AVeriTeC claim-verification records into Jev Decision Index retrieval and classification requests, embedded with Qwen3.6-35B-A3B-FP8 into 2048-d vectors for post-training CLM heads.
Jev Models & Reimplementations#open-models#classification#benchmark
Community
Hugging Face dataset of Jev Decision Index-shaped state/action embeddings built from Re-DocRED. Holds 13,312 train and 1,517 validation pairs embedded with Qwen3.6-35B-A3B-FP8 at 2048 dimensions for post-training CLM heads.
Jev Models & Reimplementations#open-models#classification#benchmark
Community
Hugging Face dataset of Jev Decision Index-shaped state/action embeddings built from TimeQA. Holds 13,278 train and 1,460 validation pairs embedded with Qwen3.6-35B-A3B-FP8 at 2048 dimensions for post-training CLM heads.
Jev Models & Reimplementations#open-models#classification#benchmark
Community
Hugging Face dataset of Jev-shaped retrieval and classification decisions built from MM-BRIGHT, holding Qwen3.6-35B-A3B-FP8 state, action, and hard-negative embeddings for post-training CLM heads.
Jev Models & Reimplementations#open-models#benchmark#evaluation
Community
Hugging Face dataset that renders HoVer claim-verification items as Jev Decision Index retrieval and classification requests, with Qwen3.6-35B-A3B-FP8 state, action, and hard-negative embeddings for post-training CLM heads.
Jev Models & Reimplementations#classification#benchmark#open-models
Community
From 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.