VectraYX-Vision-1B: A Sub-2B Spanish/LATAM Cybersecurity Vision-Language Model with Structured Visual Reasoning and Native Tool Use
Juan S. Santillana
Abstract
25 pages, 1 figure, 10 tables. v3: transplanting a natively-trained visual tower (Qwen2-VL) onto the same frozen decoder takes the failing 8-nibble address field from 0.00 to 0.81 exact, at a coarser token budget than 2x2 tiling, refuting resolution as the operative variable. Second pre-registered field found 63% contaminated, demoted. B6/B7 tool-id remains at floor. Code/checkpoints on HF.
Create a lesson
Related papers
Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations
Leon Bergen, Usha Bhalla, Andrew Lee et al.
Reporting Practice Matters: The Impact of Reference Choice on Chest X-ray Report Evaluation
Daniel P. Jeong, Charles Q. Li, Hossein Hosseiny et al.
Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs
Zimu Xu
Beyond Outcomes: Dual-View Relational Learning for Efficient Agent Benchmarking
Xinshuai Guo, Junjie Wu, Dolly Deng et al.
How Much is a Human Right Worth? ECtHR-NPD: A Benchmark for Predicting Non-Pecuniary Damage Awards
Yanyi Pu, Damian A. Gonzalez-Salzberg, Zheng Yuan et al.
Structured Claim-Level Discourse Representations for Dense Health Narratives
Farnoushsadat Nilizadeh, Elham Pourabbas Vafa, Shirin Nilizadeh et al.