Jev directory
CommunityJev Models & Reimplementations
Jev Boost re6-hover-averitec Embeddings (Qwen3.6-35B-A3B-FP8)
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.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- 2,992 train and 339 validation state/action pairs, each carrying one hard negative (k=1).
- Vectors are 2048-d float16, L2-normalized, last-token pooled through vLLM.
- States render the benchmark request plus its own instructions and never list the options.
- One-token actions and negatives are zeroed and resolved from a separate single-token embedding table.
- AVeriTeC source text is CC BY-NC 4.0; the repack card carries no license of its own.
Questions about this entry
Short answers drawn from the linked project or page.
It supplies state/action embeddings for post-training CLM heads, consumed through clm_train.data.jev_boost as described in docs/jev-boost.md. Records come from AVeriTeC, converted into decisions shaped like the Jev Decision Index's Retrieval & Classification requests for HoVer claim verification.
The train split holds 2,992 pairs and the validation split holds 339 pairs, each with one hard negative. Metadata reports num_samples 2,992, hidden_size 2,048, max_model_len 16,384 and state_head_tokens 2,048.
Qwen/Qwen3.6-35B-A3B-FP8 produced them using vLLM pooling with the last token, L2-normalized, at 2048 dimensions. The renderer is clm_train.render.Renderer and the multimodal flag is false.
Each split contains state_embeddings.pt, action_embeddings.pt, neg_embeddings.pt, neg_counts.pt, single_tokens.npz and metadata.json. Embedding tensors are float16, negatives are zero-padded, and neg_counts.pt is int64. Total file size is 42.6 MB.
vLLM's pooling of a one-token prompt is noise for this model, so one-token sequences are never embedded. They are zeroed here and single_tokens.npz points them at jackyk02/qwen3.6-35b-a3b-fp8-single-token-embeddings, which clm_train.data.stores.EmbeddingDir applies on load.
The repack's card carries no license; the source's license applies, and AVeriTeC is CC BY-NC 4.0. The vectors are a derivative of that source text, so the non-commercial terms of the original dataset carry over to this repack.
Items whose state shares a paragraph with a Jev suite item are dropped, and at most 5,000 rows are kept as a hash sample. Hard negatives are the other options, at most 8 per row, with near-copies of the gold option removed.
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 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
Hugging Face dataset of Qwen3.6-35B-A3B-FP8 embeddings over PRESTO's English tool-use requests, shaped as Jev Decision Index When2Call multiple-choice items for post-training CLM heads.
Jev Models & Reimplementations#open-models#benchmark#classification
Community
A family of 0.8B-9B ophthalmic decision models that score candidate answers without autoregressive decoding, trained on RetiJev-Data: 78,158 clinical-decision and diagnosis questions and 50,074 guideline-grading questions. RetiJev-9B answers a case in 158 ms on one GPU.
Jev Models & ReimplementationsArticle#arxiv#open-models#benchmark
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.