English
Related papers

Related papers: Instruction Tuning Changes How Upstream State Cond…

200 papers

When training large language models (LLMs), it is common practice to track downstream task performance throughout the training process and select the checkpoint with the highest validation score. However, downstream metrics often exhibit…

Computation and Language · Computer Science 2025-10-07 Yuto Nishida , Masaru Isonuma , Yusuke Oda

We localize the policy routing mechanism in alignment-trained language models. An intermediate-layer attention gate reads detected content and triggers deeper amplifier heads that boost the signal toward refusal. In smaller models the gate…

Computation and Language · Computer Science 2026-05-04 Gregory N. Frank

Modern language models can imitate complex patterns through few-shot learning, enabling them to complete challenging tasks without fine-tuning. However, imitation can also lead models to reproduce inaccuracies or harmful content if present…

Machine Learning · Computer Science 2024-03-13 Danny Halawi , Jean-Stanislas Denain , Jacob Steinhardt

Role prompts of the form As X, do Y admit a clean linear decomposition at one specific site in the residual stream: the prompt-to-answer transition -- the last prompt token together with the first two generated tokens -- in an early/mid…

Computation and Language · Computer Science 2026-05-25 Eric Xu

Preference learning is a widely adopted post-training technique that aligns large language models (LLMs) to human preferences and improves specific downstream task capabilities. In this work we systematically investigate how specific…

Computation and Language · Computer Science 2024-12-23 Joongwon Kim , Anirudh Goyal , Aston Zhang , Bo Xiong , Rui Hou , Melanie Kambadur , Dhruv Mahajan , Hannaneh Hajishirzi , Liang Tan

Scaling laws provide important insights that can guide the design of large language models (LLMs). Existing work has primarily focused on studying scaling laws for pretraining (upstream) loss. However, in transfer learning settings, in…

Computation and Language · Computer Science 2026-01-30 Berivan Isik , Natalia Ponomareva , Hussein Hazimeh , Dimitris Paparas , Sergei Vassilvitskii , Sanmi Koyejo

Backdoor data poisoning, inserted within instruction examples used to fine-tune a foundation Large Language Model (LLM) for downstream tasks (\textit{e.g.,} sentiment prediction), is a serious security concern due to the evasive nature of…

Cryptography and Security · Computer Science 2024-08-23 Jayaram Raghuram , George Kesidis , David J. Miller

Machine learning algorithms often contain many hyperparameters (HPs) whose values affect the predictive performance of the induced models in intricate ways. Due to the high number of possibilities for these HP configurations and their…

Prompt-Tuning is an efficient method for adapting pre-trained language models to new tasks with minimal computational overhead by modifying prompt embeddings. In this work, we investigate how crucial the phenomenon of embedding collapse,…

Computation and Language · Computer Science 2026-03-10 Sergey Sedov , Sumanth Bharadwaj Hachalli Karanam , Venu Gopal Kadamba

Language model post-training has enhanced instruction-following and performance on many downstream tasks, but also comes with an often-overlooked cost on tasks with many possible valid answers. On many tasks such as creative writing,…

Computation and Language · Computer Science 2026-03-04 Taylor Sorensen , Benjamin Newman , Jared Moore , Chan Park , Jillian Fisher , Niloofar Mireshghallah , Liwei Jiang , Yejin Choi

Using responses generated by high-performing large language models (LLMs) for instruction tuning has become a widely adopted approach. However, the existing literature overlooks a property of LLM-generated responses: they conflate world…

Computation and Language · Computer Science 2026-04-16 Tatsuya Ichinose , Youmi Ma , Masanari Oi , Ryuto Koike , Naoaki Okazaki

Pre-trained Large Language Models (LLMs) require post-training methods such as supervised fine-tuning (SFT) on instruction-response pairs to enable instruction following. However, this process can potentially harm existing capabilities…

Computation and Language · Computer Science 2025-02-04 Yihan Wang , Andrew Bai , Nanyun Peng , Cho-Jui Hsieh

We study Latent Recurrent Transformer (LRT), a lightweight augmentation of autoregressive transformers that reuses a high-level source-layer hidden state from the previous token as recurrent memory for the next token. Because this source…

Goal-conditioned policies enable decision-making models to execute diverse behaviors based on specified goals, yet their downstream performance is often highly sensitive to the choice of instructions or prompts. To bypass the limitations of…

Artificial Intelligence · Computer Science 2026-05-04 Guangyu Zhao , Kewei Lian , Haoxuan Ru , Borong Zhang , Haowei Lin , Zhancun Mu , Haobo Fu , Qiang Fu , Shaofei Cai , Zihao Wang , Yitao Liang

Supervised fine-tuning (SFT) is a common first stage of LLM post-training, teaching the model to follow instructions and shaping its behavior as a helpful assistant. At the same time, SFT may harm the fundamental capabilities of an LLM,…

Machine Learning · Computer Science 2026-04-16 Mark Rofin , Aditya Varre , Nicolas Flammarion

Large language models are pre-trained on ever-growing token budgets under the assumption that better pre-training performance translates to improved downstream models. In this work, we challenge this assumption and show that extended…

Computation and Language · Computer Science 2025-03-31 Jacob Mitchell Springer , Sachin Goyal , Kaiyue Wen , Tanishq Kumar , Xiang Yue , Sadhika Malladi , Graham Neubig , Aditi Raghunathan

Large language models (LLMs) consistently benefit from further fine-tuning on various tasks. However, we observe that directly tuning the Instruct (i.e., instruction-tuned) models often leads to marginal improvements and even performance…

Computation and Language · Computer Science 2025-09-29 Taiqiang Wu , Runming Yang , Jiayi Li , Pengfei Hu , Yik-Chung Wu , Ngai Wong , Yujiu Yang

Backdoor attacks inject poisoning samples during training, with the goal of forcing a machine learning model to output an attacker-chosen class when presented a specific trigger at test time. Although backdoor attacks have been demonstrated…

Fine-tuned language models have been shown to exhibit biases against protected groups in a host of modeling tasks such as text classification and coreference resolution. Previous works focus on detecting these biases, reducing bias in data…

Computation and Language · Computer Science 2021-04-13 Xisen Jin , Francesco Barbieri , Brendan Kennedy , Aida Mostafazadeh Davani , Leonardo Neves , Xiang Ren

We show that continual pretraining on plausible misinformation can overwrite specific factual knowledge in large language models without degrading overall performance. Unlike prior poisoning work under static pretraining, we study repeated…

Machine Learning · Computer Science 2026-02-09 Svetlana Churina , Niranjan Chebrolu , Kokil Jaidka
‹ Prev 1 4 5 6 7 8 10 Next ›