AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

AMiD: Knowledge Distillation for LLMs with $\alpha$-mixture Assistant Distribution

Donghyeok Shin, Yeongmin Kim, Suhyeon Jo, Byeonghu Na, Il-chul Moon

ICLR 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace SRKL/TAID/ABKD with AMiD alpha-mixture assistantGPT-2 XL -> GPT-2 0.1B · instruction-following avgROUGE-L: 21.76 -> 23.40 (vs ABKD)Improved
AMiD vs ABKD on larger studentGPT-2 XL -> GPT-2 Large · instruction-following avgROUGE-L: 24.88 -> 25.84Improved
use small (negative) alpha with D_KL assistant (ablation)AMiD, GPT-2 0.1B student · instruction-following avgROUGE-L: 18.16 (alpha=1.0) -> 22.66 (alpha=-5.0)Improved
add assistant distribution to KL (ablation)GPT-2 0.1B student · instruction-following avgROUGE-L: 17.61 -> 22.66Improved
use alpha=1.0 with D_RKL assistant (ablation)AMiD, GPT-2 0.1B student · instruction-following avgROUGE-L: 21.46 -> 3.94Hurt
use alpha=1 (e-mixture) with D_AB (ablation)AMiD, GPT-2 0.1B student · instruction-following avgROUGE-L: 21.76 -> 15.92Hurt
increase alpha (ablation)AMiD · quality-diversity (ROUGE-L, Self-BLEU)quality vs diversityMixed
raise lambda to 0.9 (ablation)AMiD D_AB · instruction-following avgROUGE-LHurt
use lambda 0.5 instead of 0.1 (ablation)AMiD D_AB · instruction-following avgROUGE-LImproved
use alpha != +-1 instead of fixed alpha=-1 or 1 (ablation)AMiD with fixed data · instruction-following avgROUGE-L: 19.32 (no assistant) -> 21.88 (Fixed data)Improved
assistant with alpha=-1 on Gemma-2B translation and GSM8K (ablation)AMiD, Gemma-7B -> 2B · Translation, GSM8KCOMET / accuracy: 74.21 -> 52.83 COMET; 24.26 -> 00.00 accHurt
swap m-mixture alpha=-1 for alpha=-5 in DistiLLM-2DistiLLM-2, Qwen2.5-7B -> 1.5B · AlpacaEval, Evol-Inst, UltraFeed avgwin rate: 70.4 -> 71.7Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy