DH{}DataHunter

Dataset from Hugging Face · Salesforce

LALM Judge Validation on Full-Duplex Voice Agents

Open on Hugging Face
License
CC BY 4.0commercial use allowedLicense text
This license allows commercial use. Usually you must credit the author; read the license for the exact terms.
Formats
CSV, CFF, MD, PNG, PY
Size
981 kB
Updated
Jul 23, 2026

Description

LALM Judge Validation on Full-Duplex Voice Agents Companion dataset for the paper A Reliability Assessment of LALM Audio Judges for Full-Duplex Voice Agents. This repository contains the anonymised ratings, adversarial-defect recall tables, JSON schemas, and analysis scripts used to produce every headline number, table, and figure in that paper. Summary 209 rated stereo sessions: 152 full-duplex agent-client conversations across 13 accent-and-condition strata…

Description from Hugging Face, shortened.

Files (27)

  • CFFCITATION.cff1,004 byte
  • MDCODE_OF_CONDUCT.md5 kB
  • MDCONTRIBUTING.md2.3 kB
  • —LICENSE12 kB
  • MDSECURITY.md452 byte
  • CSVdata/cross_model_comparison.csv1.2 kB
  • CSVdata/d_personality_vs_codec.csv625 byte
  • CSVdata/e_ceiling_collapse.csv1.6 kB
  • CSVdata/full_per_session.csv133 kB
  • CSVdata/h1_per_dim.csv1.2 kB
  • CSVdata/h1_per_dim_gemini31propreview.csv640 byte
  • CSVdata/h1_per_dim_gemini35flash.csv639 byte
  • CSVdata/h1_per_dim_natural.csv516 byte
  • CSVdata/h2_per_config_per_dim.csv6 kB
  • CSVdata/h3_defect_recall_3rater.csv2.7 kB
  • CSVdata/h3_defect_recall_with_ci.csv4.3 kB
  • CSVdata/lalm_as_4th_rater.csv876 byte
  • CSVdata/rho_hh_ci.csv528 byte
  • PNGfigures/fig1_three_regime_taxonomy.png150 kB
  • PNGfigures/fig2_lalm_as_4th_rater.png75 kB
  • PNGfigures/fig3_defect_detection_forest.png340 kB
  • PNGfigures/fig4_personality_vs_codec.png112 kB
  • PNGfigures/fig5_ceiling_partition.png119 kB
  • MDscripts/README.md1.4 kB
  • PYscripts/compute_agreement.py4 kB
  • PYscripts/compute_krippendorff.py2.7 kB
  • PYscripts/compute_newcombe_wilson.py3.9 kB

DataHunter does not store or host these files: every link leads to the original source. Check the license there before using the data.

Search other datasets