m-polignano/decision4rec-kev-9b

Decision4Rec Kev-9B (pool reader)

The trained decision model of Decision4Rec-Full at 9B parameters, from Decision4Rec: it reads the 30 best candidates of a multi-agent pool, with item text, statements of the knowledge graph of the scenario and the agents that found each candidate, and returns one probability per candidate in a single forward pass (System One protocol, choice question).

Model details

Base model	Qwen/Qwen3.5-9B-Base (revision 68c46c4b3498877f3ef123c856ecfde50c39f404), frozen
Starting point	the released jaredpalmer/kev-9b adapter and pointer head
Trained parameters	rank-16 LoRA adapters on all linear layers, pointer head
Training data	a fixed random 8,000 (seed 2026) of the 34,667 real next choices used for the 0.8B and 4B models (Amazon Video Games, Amazon Musical Instruments, Steam), over the 30 best candidates of the pool; the same 8,000 records train the data-matched 0.8B, 4B, 9B and 27B models of the paper
Objective	likelihood of the item the user really chose next; no outputs of any commercial model were used
Recipe	one epoch, learning rate 5e-5, one-cycle schedule, micro-batch 2 x 4 accumulation (8 records per step), bf16, gradient checkpointing
Trainer	Kev kev.train, commit 84847f0, on one NVIDIA DGX Spark

Results

NDCG@10 of Decision4Rec-Full on 2,000 development test users per domain (full-catalogue ranking, leave-one-out). Industrial and Scientific was never seen in training.

System	Video Games	Musical Instr.	Steam	Industrial & Sci.
best trained baseline	0.0622	0.0370	0.0790	0.0310
pool and residual ranker, no decision model	0.0730	0.0468	0.0786	0.0386
with this model (fused)	0.0736	0.0485	0.0798	0.0406

Use

cd ~/kev && uv run --extra serve python -m kev.serve --run m-polignano/decision4rec-kev-9B --port 8009
decision4rec run --config configs/small/vg_rtm.yaml --backend configs/backends/kev-9B-rtm8k.yaml

Limitations

• Trained on English product and game catalogues; dense domains with long histories (MovieLens) gain less.
• Development results only; a confirmatory test on fresh users is planned.
• It ranks candidates it is given; it does not retrieve.

Licenses

The adapter and head are released under Apache-2.0. The base model keeps its own license. The training data come from public research datasets that are not redistributed here.

Citation

@article{polignano2026decision4rec,
  title  = {Decision4Rec: A Jev-Inspired Framework of Decision Models on Knowledge and Active Graphs for Multi-Agent Recommender Systems},
  author = {Polignano, Marco and Mehta, Het Darshan and Lops, Pasquale and De Luca, Ernesto William and Semeraro, Giovanni},
  year   = {2026}
}