{
"cells": [
{
"cell_type": "markdown",
"id": "64ca5797",
"metadata": {},
"source": [
"# GLiNER2 β el modelo unico para `graph_explorer`\n",
"\n",
"Tras descartar GLiREL (notebook 02) y aceptar mREBEL con caveat de licencia (notebook 03), encontramos **`fastino/gliner2-large-v1`**: NER + RE en un solo modelo, **Apache 2.0**, soporta castellano nativo, **20-30Γ mas rapido** que mREBEL.\n",
"\n",
"| | GLiNER + GLiREL | GLiNER + mREBEL | **GLiNER2** |\n",
"|---|---|---|---|\n",
"| Modelos | 2 | 2 | **1** |\n",
"| TamaΓ±o total | 2.1 GB | 3.0 GB | **0.7 GB** |\n",
"| Latencia 8 frases ES | 1.0s | 25s | **1.2s** |\n",
"| Latencia 30 frases ES | ~3s | ~90s | **4.2s** |\n",
"| Calidad ES corporate | 1 falsa | 4/5 OK | **5-6/8 OK** |\n",
"| Calidad ES OSINT | sin probar | sin probar | **funciona** |\n",
"| Licencia | Apache 2.0 | CC BY-NC-SA 4.0 | **Apache 2.0** |\n",
"| Idioma | EN-centric | 18 idiomas | EN/ES/FR |\n",
"\n",
"Este notebook empotra los datos del benchmark v2 (`benchmark_v2.json`) y construye el grafo final."
]
},
{
"cell_type": "markdown",
"id": "1ec3450e",
"metadata": {},
"source": [
"## 1. Setup"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "ac1a949e",
"metadata": {
"execution": {
"iopub.execute_input": "2026-05-04T13:42:13.793562Z",
"iopub.status.busy": "2026-05-04T13:42:13.793426Z",
"iopub.status.idle": "2026-05-04T13:42:18.411410Z",
"shell.execute_reply": "2026-05-04T13:42:18.410428Z"
}
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[0;93m2026-05-04 15:42:15.263782788 [W:onnxruntime:Default, device_discovery.cc:283 GetGpuDevices] Failed to detect devices under \"/sys/class/drm/card0\": device_discovery.cc:93 ReadFileContents Failed to open file: \"/sys/class/drm/card0/device/vendor\"\u001b[m\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"corpora benchmarked: ['es_corporate_short', 'es_corporate_long', 'es_osint', 'en_corporate_short']\n"
]
}
],
"source": [
"import os, sys, json, warnings, time\n",
"warnings.filterwarnings('ignore')\n",
"os.environ.setdefault('HF_HUB_DISABLE_PROGRESS_BARS', '1')\n",
"from pathlib import Path\n",
"\n",
"_pf = '/home/lucas/fn_registry/python/functions'\n",
"sys.path = [p for p in sys.path if not p.startswith(_pf + '/')]\n",
"if _pf not in sys.path: sys.path.insert(0, _pf)\n",
"\n",
"import pandas as pd\n",
"import networkx as nx\n",
"import matplotlib.pyplot as plt\n",
"from gliner2 import GLiNER2\n",
"\n",
"BENCH = json.loads(Path('../benchmark_v2.json').read_text())\n",
"print('corpora benchmarked:', list(BENCH.keys()))"
]
},
{
"cell_type": "markdown",
"id": "8ad929f7",
"metadata": {},
"source": [
"## 2. Cargar GLiNER2 (warm β modelo cacheado)"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "998dd198",
"metadata": {
"execution": {
"iopub.execute_input": "2026-05-04T13:42:18.413727Z",
"iopub.status.busy": "2026-05-04T13:42:18.413052Z",
"iopub.status.idle": "2026-05-04T13:42:31.934909Z",
"shell.execute_reply": "2026-05-04T13:42:31.934026Z"
}
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"You are using a model of type extractor to instantiate a model of type . This is not supported for all configurations of models and can yield errors.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"============================================================\n",
"π§ Model Configuration\n",
"============================================================\n",
"Encoder model : microsoft/deberta-v3-large\n",
"Counting layer : count_lstm\n",
"Token pooling : first\n",
"============================================================\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"GLiNER2 ready in 13.5s\n"
]
}
],
"source": [
"t0 = time.time()\n",
"model = GLiNER2.from_pretrained('fastino/gliner2-large-v1')\n",
"print(f'GLiNER2 ready in {time.time()-t0:.1f}s')"
]
},
{
"cell_type": "markdown",
"id": "101c5829",
"metadata": {},
"source": [
"## 3. Resumen del benchmark sobre 4 corpora\n",
"\n",
"Datos de `run_benchmark_v2.py` corrido el 2026-05-04. Cada fila es una pasada GLiNER2 con su schema (entities + relations) sobre el corpus."
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "a3673e52",
"metadata": {
"execution": {
"iopub.execute_input": "2026-05-04T13:42:31.936688Z",
"iopub.status.busy": "2026-05-04T13:42:31.936494Z",
"iopub.status.idle": "2026-05-04T13:42:31.947737Z",
"shell.execute_reply": "2026-05-04T13:42:31.946991Z"
}
},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" corpus | \n",
" chars | \n",
" words | \n",
" time_s | \n",
" ents | \n",
" rels | \n",
" rels/word | \n",
"
\n",
" \n",
" \n",
" \n",
" | 0 | \n",
" es_corporate_short | \n",
" 658 | \n",
" 104 | \n",
" 1.185 | \n",
" 14 | \n",
" 8 | \n",
" 0.0769 | \n",
"
\n",
" \n",
" | 1 | \n",
" es_corporate_long | \n",
" 2582 | \n",
" 400 | \n",
" 4.212 | \n",
" 60 | \n",
" 6 | \n",
" 0.0150 | \n",
"
\n",
" \n",
" | 2 | \n",
" es_osint | \n",
" 724 | \n",
" 98 | \n",
" 1.071 | \n",
" 11 | \n",
" 5 | \n",
" 0.0510 | \n",
"
\n",
" \n",
" | 3 | \n",
" en_corporate_short | \n",
" 314 | \n",
" 49 | \n",
" 0.767 | \n",
" 9 | \n",
" 9 | \n",
" 0.1837 | \n",
"
\n",
" \n",
"
\n",
"