Chip is the main NL2Hull ship-design decision model. It adapts a Kev-style typed decision interface to Chinese natural-language ship-form editing requests and predicts probability distributions over ordered editing decisions. This release contains the LoRA adapter and the learned decision head for the 0.8-billion-parameter Chip model. The adapter is applied to Qwen/Qwen3.5-0.8B-Base; the head.pt file is required by the NL2Hull/Kev inference implementation. - adaptermodel.safetensors: LoRA adapter weights - adapterconfig.json: PEFT adapter configuration - head.pt: learned typed-decision head - tokenizer.json, tokenizerconfig.json, chattemplate.jinja: tokenizer files - trainingconfig.json…
Independent publisher
Wenhua Huo
wenhuahuo
AI4CFD, AI4CAD
Models in Library2
Datasets in Library1
Models on Hugging Face2
Followers—
Compare the published evidence
This profile connects the 3 model and dataset records shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Published tasks: Text generation.
Reported licenses: mit. Open a record to review its specific license and conditions.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.
Models
Chip is the NL2Hull ship-design decision model family. It adapts a Kev-style typed decision interface to Chinese natural-language ship-form editing requests and predicts probability distributions over ordered editing decisions. The released 0.8-billion-parameter adapter is available at wenhuahuo/chip-0.8b. The model is designed to be used with the NL2Hull repository and the Ship Design Decision Dataset. The model and evaluation protocol are described in NL2Hull: A Natural Language-Driven Constrained Ship Design Decision Framework.
Datasets
The Ship Design Decision Dataset (SDD Dataset) contains cleaned Chinese natural-language requests paired with typed decisions for constrained ship-form editing. SDDBench is the fixed 5,000-record evaluation subset used in the NL2Hull paper. - train.jsonl: 88,604 records - validation.jsonl: 22,758 records - test.jsonl: 23,196 records - test5k.jsonl: frozen 5,000-record benchmark subset - languagesamples.jsonl: source language snapshot The split policy separates hull families across training, validation, and test sets. The training split uses DTC, DTMB 5415, KCS, KVLCC2, containership, frigate, and NPL variants; validation uses S-175 and Series 60; test uses Wigley and workboat hulls. Each…