license: apache-2.0
language:
- zh
- en
pretty_name: ScholarSearchAgent Auxiliary Models
tags:
- information-retrieval
- reranking
- text-classification
- lora
- academic-search
ScholarSearchAgent Auxiliary Models
This dataset repository hosts the trained auxiliary model artifacts used by ScholarSearchAgent, an intelligent academic paper search and recommendation system for complex research queries.
Source code: https://github.com/Kaedeser/ScholarSearchAgent
Download And Restore
The complete model bundle is stored as 200 MiB archive parts named
ScholarSearchAgentAuxiliaryModel_20260723.tar.part0001,
ScholarSearchAgentAuxiliaryModel_20260723.tar.part0002, and so on. Download
all parts in numerical order, concatenate them, verify the SHA-256 value, then
extract the resulting TAR archive.
PowerShell:
Get-ChildItem "ScholarSearchAgentAuxiliaryModel_20260723.tar.part*" |
Sort-Object Name |
Get-Content -AsByteStream |
Set-Content -AsByteStream "ScholarSearchAgentAuxiliaryModel_20260723.tar"
Get-FileHash "ScholarSearchAgentAuxiliaryModel_20260723.tar" -Algorithm SHA256
tar -xf "ScholarSearchAgentAuxiliaryModel_20260723.tar"
Expected SHA-256:
c15830db9b3c126d584b1ffbdef17cd70a4101d20470ad78661eb883b4fe4a46
Included Models
| Directory after extraction | Purpose | Model form | Reported result |
| --- | --- | --- |
| models/query_gate_biobert | Detect whether a query should enter the academic-search pipeline | BioBERT sequence classifier | Test accuracy: 0.997143 |
| models/intent_biobert | Classify academic search intent | BioBERT sequence classifier | See packaged training metadata |
| models/selector_reranker | Rerank recalled paper candidates | Sentence Transformers CrossEncoder | F1: 88.81% |
| models/crawler_strategy_lora | Select section-expansion strategies | Qwen2.5-3B-Instruct LoRA adapter | Section F1: 0.3007 |
The crawler strategy artifact is a LoRA adapter. The public
Qwen2.5-3B-Instruct base model is not redistributed in this repository and
must be obtained separately under its original license.
Data And Privacy
The archive contains model weights, tokenizer/configuration files, and training metadata required for inference. It does not contain API keys, passwords, private endpoints, raw training datasets, source-code history, or participant identity information.