# FIRST: Teach A Reliable Large Language Model Through Efficient Trustworthy Distillation

arXiv: https://arxiv.org/abs/2408.12168 · HTML: https://mlsanity.com/paper/2408.12168/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 19 references that are in the map.
- First posted on arXiv 22 Aug 2024.

## Builds on (most cited first, 11 of 11)

- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [When Does Label Smoothing Help?](https://mlsanity.com/paper/1906.02629/) (2019)
- [How to Fine-Tune BERT for Text Classification?](https://mlsanity.com/paper/1905.05583/) (2019)
- [Fine-Tuning Pretrained Language Models: Weight Initializations, Data Orders, and Early Stopping](https://mlsanity.com/paper/2002.06305/) (2020)
- [LanguageMPC: Large Language Models as Decision Makers for Autonomous Driving](https://mlsanity.com/paper/2310.03026/) (2023)
- [DistillSpec: Improving Speculative Decoding via Knowledge Distillation](https://mlsanity.com/paper/2310.08461/) (2023)
- [Symbolic Chain-of-Thought Distillation: Small Models Can Also "Think" Step-by-Step](https://mlsanity.com/paper/2306.14050/) (2023)
- [Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Data](https://mlsanity.com/paper/2302.12822/) (2023)
- [LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models](https://mlsanity.com/paper/2306.12420/) (2023)
- [How Trustworthy are Open-Source LLMs? An Assessment under Malicious Demonstrations Shows their Vulnerabilities](https://mlsanity.com/paper/2311.09447/) (2023)

## Built on by (most cited first, 11 of 11)

- [Entropy-Aware On-Policy Distillation of Language Models](https://mlsanity.com/paper/2603.07079/) (2026)
- [Knowledge Distillation and Dataset Distillation of Large Language Models: Emerging Trends, Challenges, and Future Directions](https://mlsanity.com/paper/2504.14772/) (2025)
- [CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization](https://mlsanity.com/paper/2510.17847/) (2025)
- [Adapt-Pruner: Adaptive Structural Pruning for Efficient Small Language Model Training](https://mlsanity.com/paper/2502.03460/) (2025)
- [Zero Token-Driven Deep Thinking in LLMs: Unlocking the Full Potential of Existing Parameters via Cyclic Refinement](https://mlsanity.com/paper/2502.12214/) (2025)
- [Bridge-Coder: Unlocking LLMs' Potential to Overcome Language Gaps in Low-Resource Code](https://mlsanity.com/paper/2410.18957/) (2024)
- [UNDO: Understanding Distillation as Optimization](https://mlsanity.com/paper/2504.02521/) (2025)
- [FOCUS & RePAIR: Mitigating Text Degeneration via Token-Level Guidance for Pruned Large Language Models](https://mlsanity.com/paper/2608.26676/) (2026)
- [Persistent Teacher Anchoring for Tool-Using Agents](https://mlsanity.com/paper/2609.04773/) (2026)
- [ACLArena: Agent Continue Learning in Multi-stage Post-training](https://mlsanity.com/paper/2609.23989/) (2026)
- [ReTaCo: Residual-Target Control for On-Policy Distillation](https://mlsanity.com/paper/2609.39275/) (2026)

## Abstract (arXiv)

Large language models (LLMs) have become increasingly prevalent in our daily lives, leading to an expectation for LLMs to be trustworthy -- - both accurate and well-calibrated (the prediction confidence should align with its ground truth correctness likelihood). Nowadays, fine-tuning has become the most popular method for adapting a model to practical usage by significantly increasing accuracy on downstream tasks. Despite the great accuracy it achieves, we found fine-tuning is still far away from satisfactory trustworthiness due to "tuning-induced mis-calibration". In this paper, we delve deeply into why and how mis-calibration exists in fine-tuned models, and how distillation can alleviate the issue. Then we further propose a brand new method named Efficient Trustworthy Distillation (FIRST), which utilizes a small portion of teacher's knowledge to obtain a reliable language model in a cost-efficient way. Specifically, we identify the "concentrated knowledge" phenomenon during distillation, which can significantly reduce the computational burden. Then we apply a "trustworthy maximization" process to optimize the utilization of this small portion of concentrated knowledge before transferring it to the student. Experimental results demonstrate the effectiveness of our method, where better accuracy (+2.3%) and less mis-calibration (-10%) are achieved on average across both in-domain and out-of-domain scenarios, indicating better trustworthiness.