ciftselcuk's picture
Initial deployment: GLiNER Large entity extractor with 70 labels
f0b69ef

A newer version of the Gradio SDK is available: 6.1.0

Upgrade
metadata
title: T9 Oracle Entity Extractor
emoji: 🔬
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
hardware: t4-small

T9 Oracle Entity Extractor

Zero-shot NER for Medical Device Technical Documentation

This Space provides entity extraction using GLiNER Large (1.7GB) for technical documentation in the medical device domain.

Features

  • 70 Entity Labels across 10 tiers
  • Zero-shot learning - no training required
  • Medical device focus - optimized for endoscope equipment, parts, specifications
  • Hardware: NVIDIA T4 GPU for fast inference

Entity Types

Tier 1: Critical Identifiers

  • part_number, component_name, manufacturer, model_number

Tier 2: Specifications

  • pressure, temperature, voltage, current, material, dimensions, flow_rate, power

Tier 3: Standards & Compliance

  • standard_reference (ISO, ASTM, EN, IEC, ANSI), certification, compliance

Tier 4-10: Additional Labels

  • Thread standards, geometry, documentation, operational parameters, manufacturing IDs, medical device specific, visual elements, quality & maintenance

API Usage

from gradio_client import Client

client = Client("YOUR_USERNAME/t9-oracle-gliner-entity-extractor")

text = "Part Number: A70002-2, Material: SS316L, Pressure: 60 psi"
result = client.predict(text, api_name="/extract")
print(result)

Configuration

  • Model: urchade/gliner_large-v2.1
  • GPU: NVIDIA T4 (16GB VRAM)
  • Cost: $0.60/hour (Persistent)
  • Max input: 10,000 characters per request

Project

Part of the T9 Oracle Knowledge Base Extraction System for Auto Sink medical device documentation.