GeoAgenticRag JEV
Final merged Qwen3.5-0.8B weights fine-tuned for structured geospatial agent decisions.
File
model.safetensors— final merged model weights (about 1.75 GB).
This repository contains the final merged weights only. It does not include intermediate training checkpoints or adapters. The checkpoint was selected at epoch 2. Training used simulated states and rule-generated labels.
Base model: Qwen/Qwen3.5-0.8B. License: Apache-2.0.
The matching model configuration and tokenizer are not included in this minimal weights repository. Use the base model files and the GeoAgenticRag model configuration when loading these weights.
Evaluation snapshot
On an exploratory 100-question MapQA Illinois sample (50 MapQA-ADJ and 50
MapQA-AME), the paired trained-JEV configuration completed 95 questions and
scored Recall@1/5/10 of 64.47/74.74/75.26%, MRR 72.27%, NDCG@10 72.61%, and
Hit@10 76.84%. Metrics use completed questions only; five cases failed. This
configuration pairs the trained JEV with the base reranker and uses
max_replans=0. The sample is a component study, not the four-dataset
benchmark result.