metadata
base_model: LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct
language:
- ko
- en
license: cc-by-nc-4.0
tags:
- speech-to-sft
- emnlp-2026-industry
- anonymous
emnlp2026_expertqa_exaone3.5-7.8b_exp1_1_medical_seed2
SFT checkpoint from the EMNLP 2026 Industry Track submission A Factorial Ablation of a Speech-to-SFT Pipeline: Differential Effects on Data Quality and Downstream Transfer (anonymous).
| Field | Value |
|---|---|
| Pipeline condition | Exp 1-1 (Phase 0 only — transcript refinement) |
| Domain | medical |
| Seed | 2 |
| Base model | LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct |
| Training | LoRA (rank 16, α=32, QLoRA 4-bit, lr 2e-4, 3 epochs) |
| Upstream STT | In-house STT (paper main pipeline) |
| License | CC BY-NC 4.0 (research and non-commercial use only) |
Intended use: research and non-commercial use only, matching the consent scope of the source audio.
Companion repository (code, configs, prompts, sample QA): https://anonymous.4open.science/r/anon-emnlp2026-expertqa-7616