The prevalence and strong capability of large language models (LLMs) present significant safety and ethical risks if exploited by malicious users. To prevent the potentially deceptive usage of LLMs, recent work has proposed algorithms to detect LLM-generated text and protect LLMs. In this paper, we investigate the robustness and reliability of these LLM detectors under adversarial attacks. We study two types of attack strategies: 1) replacing certain words in an LLM’s output with their synonyms given the context; 2) automatically searching for an instructional prompt to alter the writing style of the generation. In both strategies, we leverage an auxiliary LLM to generate the word replacements or the instructional prompt. Different from previous works, we consider a challenging setting where the auxiliary LLM can also be protected by a detector. Experiments reveal that our attacks effectively compromise the performance of all detectors in the study with plausible generations, underscoring the urgent need to improve the robustness of LLM-generated text detection systems. Code is available at https://github.com/shizhouxing/LLM-Detector-Robustness
more »
« less
This content will become publicly available on August 4, 2025
CacheGen: KV Cache Compression and Streaming for Fast Large Language Model Serving
As large language models (LLMs) take on complex tasks, their inputs are supplemented with longer contexts that incorporate domain knowledge. Yet using long contexts is challenging as nothing can be generated until the whole context is processed by the LLM. While the context-processing delay can be reduced by reusing the KV cache of a context across different inputs, fetching the KV cache, which contains large tensors, over the network can cause high extra network delays.
CacheGen is a fast context-loading module for LLM systems. First, CacheGen uses a custom tensor encoder, leveraging KV cache's distributional properties to encode a KV cache into more compact bitstream representations with negligible decoding overhead, to save bandwidth usage. Second, CacheGen adapts the compression level of different parts of a KV cache to cope with changes in available bandwidth, in order to maintain low context-loading delay and high generation quality. We test CacheGen on popular LLMs and datasets. Compared to the recent systems that reuse the KV cache, CacheGen reduces the KV cache size by 3.5--4.3x and the total delay in fetching and processing contexts by 3.2--3.7x with negligible impact on the LLM response quality. Our code is at: https://github.com/UChi-JCL/CacheGen.
more »
« less
- Award ID(s):
- 2313190
- PAR ID:
- 10536862
- Publisher / Repository:
- Association for Computing Machinery, New York, NY, United States
- Date Published:
- ISSN:
- 0146-4833
- ISBN:
- 979-8-4007-0614-1
- Format(s):
- Medium: X
- Location:
- Sydney NSW Australia
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Practitioners frequently take multiple samples from large language models (LLMs) to explore the distribution of completions induced by a given prompt. While individual samples can give high-quality results for given tasks, collectively there are no guarantees of the distribution over these samples induced by the generating LLM. In this paper, we empirically evaluate LLMs’ capabilities as distribution samplers. We identify core concepts and metrics underlying LLM-based sampling, including different sampling methodologies and prompting strategies. Using a set of controlled domains we evaluate the error and variance of the distributions induced by the LLM. We find that LLMs struggle to induce reasonable distributions over generated elements, suggesting that practitioners should more carefully consider the semantics and methodologies of sampling from LLMs.more » « less
-
High-quality knowledge graphs (KGs) play a crucial role in many applications. However, KGs created by automated information extraction systems can suffer from erroneous extractions or be inconsistent with provenance/source text. It is important to identify and correct such problems. In this paper, we study leveraging the emergent reasoning capabilities of large language models (LLMs) to detect inconsistencies between extracted facts and their provenance. With a focus on “open” LLMs that can be run and trained locally, we find that few-shot approaches can yield an absolute performance gain of 2.5-3.4% over the state-of-the-art method with only 9% of training data. We examine the LLM architectures’ effect and show that Decoder-Only models underperform Encoder-Decoder approaches. We also explore how model size impacts performance and counterintuitively find that larger models do not result in consistent performance gains. Our detailed analyses suggest that while LLMs can improve KG consistency, the different LLM models learn different aspects of KG consistency and are sensitive to the number of entities involved.more » « less
-
more » « less
High-quality knowledge graphs (KGs) play a crucial role in many applications. However, KGs created by automated information extraction systems can suffer from erroneous extractions or be inconsistent with provenance/source text. It is important to identify and correct such problems. In this paper, we study leveraging the emergent reasoning capabilities of large language models (LLMs) to detect inconsistencies between extracted facts and their provenance. With a focus on ``open'' LLMs that can be run and trained locally, we find that few-shot approaches can yield an absolute performance gain of 2.5-3.4% over the state-of-the-art method with only 9% of training data. We examine the LLM architectures' effect and show that Decoder-Only models underperform Encoder-Decoder approaches. We also explore how model size impacts performance and counterintuitively find that larger models do not result in consistent performance gains. Our detailed analyses suggest that while LLMs can improve KG consistency, the different LLM models learn different aspects of KG consistency and are sensitive to the number of entities involved.
-
Large language models (LLMs), such as GPT-3 and GPT-4, have demonstrated exceptional performance in various natural language processing tasks and have shown the ability to solve certain reasoning problems. However, their reasoning capabilities are limited and relatively shallow, despite the application of various prompting techniques. In contrast, formal logic is adept at handling complex reasoning, but translating natural language descriptions into formal logic is a challenging task that non-experts struggle with. This paper proposes a neuro-symbolic method that combines the strengths of large language models and answer set programming. Specifically, we employ an LLM to transform natural language descriptions of logic puzzles into answer set programs. We carefully design prompts for an LLM to convert natural language descriptions into answer set programs in a step by step manner. Surprisingly, with just a few in-context learning examples, LLMs can generate reasonably complex answer set programs. The majority of errors made are relatively simple and can be easily corrected by humans, thus enabling LLMs to effectively assist in the creation of answer set programs.more » « less
