JEV-27B-VL vs mammoth-1

JEV-27B-VL

A cousin for typed decisions with probabilities over text and images, answering in one pass with optional thinking. It needs about 67 GB in BF16 on vLLM, with no AutoTrust 4-bit build or hosted API. The main risks are labels distilled from TypeSafe's Jev against its customer terms and serving through a repository script on a vLLM development build.

mammoth-1

A cousin for typed decisions, several questions about one state answered with probabilities from one prefill. It needs about 66 GB in BF16 and a pinned, patched third-party runtime. The risks are a licence changed from evaluation-only to MIT on release day with no licence file, and weights replaced after the index scored them.

JEV-27B-VL

1. Quickest start is hf download autotrust/JEV-27B-VL then bash JEV-27B-VL/serve.sh, which serves POST /v1/decide on port 8000, but it runs the repository's serve_decide.py on a September 2026 vLLM development build
2. On a released vLLM use plain vllm serve with --enable-lora --lora-modules jev-decision=JEV-27B-VL/adapter_vllm and do the card's client-side arithmetic with decision_head.json and calibration.json
3. Keep --max-num-seqs 8, the card says larger batches return wrong System 1 probabilities on vLLM's multimodal LoRA path
4. When calling /v1/completions or /v1/chat/completions yourself pass top_k 0 and top_p 1.0, or generation_config.json's top_k 20 and top_p 0.95 truncate the probabilities
5. Choices beyond 16 options use labels the head never saw in training, and the response's adaptation field says wide-labels
6. Downscale images to about 448 px to keep vision tokens and latency down
7. Treat it as AutoTrust's model, not TypeSafe's, and its request shape differs from TypeSafe's /v1/systemone

mammoth-1

1. Run bash setup.sh <workdir> in the pytorch/pytorch:2.8.0-cuda12.8-cudnn9-devel image on one CUDA GPU. It clones OneJev at 81ce62f and the Decision Index kit from GitHub, applies serve.patch and pip-installs them, so read it first
2. Load through engine.py's DecisionModel (qev DecisionEngine). Plain Transformers or vLLM run the full-attention layers causally and won't give the trained probabilities
3. Send one state and many questions per request, each choice, yes or no (noul) or score, and read a probability per option
4. Keep requests within the trained length, the report trained at up to 16,384 tokens and engine.py refuses requests over the trained limits
5. The index's scores are for revision b9f2f2cacef2ec178ed25c9ef43e987d3422a639. Pin that revision to get the scored weights, or main for the later ones
6. Text only, the vision encoder is the base's, frozen and untrained for decisions