# LettuceDetect: A Hallucination Detection Framework for RAG Applications

LettuceDetect Logo
Because even AI needs a reality check! 🥬

## TL;DR We present **LettuceDetect**, a lightweight hallucination detector for Retrieval-Augmented Generation (RAG) pipelines. It is an **encoder-based** model built on [ModernBERT](https://github.com/AnswerDotAI/ModernBERT), released under the MIT license with ready-to-use Python packages and pretrained models. - **What**: LettuceDetect is a token-level detector that flags unsupported segments in LLM answers. 🥬 - **How**: Trained on [RAGTruth](https://aclanthology.org/2024.acl-long.585/) (18k examples), leveraging ModernBERT for context lengths up to 4k tokens. 🚀 - **Why**: It addresses (1) the context-window limits in prior encoder-only models, and (2) the high compute costs of LLM-based detectors. ⚖️ - **Highlights**: - Beats prior encoder-based models (e.g., [Luna](https://aclanthology.org/2025.coling-industry.34/)) on RAGTruth. ✅ - Surpasses fine-tuned Llama-2-13B [2] at a fraction of the size, and is highly efficient at inference. ⚡️ - Entirely **open-source** with an MIT license. 🔓 **LettuceDetect** keeps your RAG framewok **fresh** by spotting **rotten** parts of your LLM's outputs. 😊 ## Quick Links - **GitHub**: [github.com/KRLabsOrg/LettuceDetect](https://github.com/KRLabsOrg/LettuceDetect) - **PyPI**: [pypi.org/project/lettucedetect](https://pypi.org/project/lettucedetect/) - **arXiv Paper**: [2502.17125](https://arxiv.org/abs/2502.17125) - **Hugging Face Models**: - [Base Model](https://huggingface.co/KRLabsOrg/lettucedect-base-modernbert-en-v1) - [Large Model](https://huggingface.co/KRLabsOrg/lettucedect-large-modernbert-en-v1) - **Streamlit Demo**: Visit our [Hugging Face Space](https://huggingface.co/spaces/KRLabsOrg/LettuceDetect) or run locally per the GitHub instructions. --- ## Get Going Install the package: ```bash pip install lettucedetect ``` Then, you can use the package as follows: ```python from lettucedetect.models.inference import HallucinationDetector # For a transformer-based approach: detector = HallucinationDetector( method="transformer", model_path="KRLabsOrg/lettucedect-base-modernbert-en-v1" ) contexts = ["France is a country in Europe. The capital of France is Paris. The population of France is 67 million.",] question = "What is the capital of France? What is the population of France?" answer = "The capital of France is Paris. The population of France is 69 million." # Get span-level predictions indicating which parts of the answer are considered hallucinated. predictions = detector.predict(context=contexts, question=question, answer=answer, output_format="spans") print("Predictions:", predictions) # Predictions: [{'start': 31, 'end': 71, 'confidence': 0.9944414496421814, 'text': ' The population of France is 69 million.'}] ``` ## Why LettuceDetect? Large Language Models (LLMs) have made considerable advancements in NLP tasks, like GPT-4 [4], the Llama-3 models [5], or Mistral [6] (and many more). Despite the success of LLMs, **hallucinations** remain a key obstacle deploying LLMs in high-stakes scenarios (such as in healthcare or legal) [7,8]. **Retrieval-Augmented Generation (RAG)** attempts to mitigate hallucinations by grounding an LLM's responses in retrieved documents, providing external knowledge that the model can reference [9]. But even though RAG is a powerful method to reduce hallucinations, LLMs still suffer from hallucinations in these settings [1]. Hallucinations are information in the output that is nonsensical, factually incorrect, or inconsistent with the retrieved context [8]. Ji et al. [10] categorizes hallucinations into: - Intrinsic Hallucinations: Stemming from the model's preexisting internal knowledge. - Extrinsic Hallucinations: Occurring when the answer conflicts with the context or references provided While RAG approaches can mitigate intrinsic hallucinations, they are not immune to extrinsic hallucinations. Sun et al. [11] showed that models tend to prioritize their intrinsic knowledge over the external context. As LLMs remain prone to hallucinations, their applications in critical domains e.g. medical or legal, can be still flawed. ### Current Solutions for Hallucination Detection Current solutions for hallucination detection can be categorized into different categories based on the approach they take: 1. **Prompt based detectors** These methods (e.g., [RAGAS](https://github.com/explodinggradients/ragas), [Trulens](https://github.com/truera/trulens), [ARES](https://github.com/stanford-futuredata/ARES)) typically leverage zero-shot or few-shot prompts to detect hallucinations. They often rely on large LLMs (like GPT-4) and employ strategies such as SelfCheckGPT [12], LM vs. LM [13], or Chainpoll [14]. While often effective, they can be computationally expensive due to repeated LLM calls. 2. **Fine-tuned LLM detectors** Large models (e.g., Llama-2, Llama-3) can be fine-tuned for hallucination detection [1,15]. This can yield high accuracy (as shown by the RAGTruth authors using Llama-2-13B or the RAG-HAT work on Llama-3-8B) but is resource-intensive to train and deploy. Inference costs also tend to be high due to their size and slower speeds. 3. **Encoder-based detectors** Models like [Luna [2]](https://aclanthology.org/2025.coling-industry.34/) rely on a BERT-style encoder (often limited to 512 tokens) for token-level classification. These methods are generally more efficient than running a full LLM at inference but are constrained by **short context windows** and attention mechanisms optimized for smaller inputs. ### ModernBERT for long context ModernBERT [3] is a drop in replacement for BERT, and is a state of the art encoder-only transformers architecture that incorporates several modern design improvements over the original BERT model such as it uses Rotary Positional Embeddings ([RoPe](https://huggingface.co/blog/designing-positional-encoding)) to handle sequences of up to 8,192 tokens, [unpadding optimization](https://arxiv.org/abs/2208.08124) to eliminate wasted computation on padding tokens, and [GeGLU](https://arxiv.org/abs/2002.05202) activation layers for enhanced expressiveness and [alternating attention](https://arxiv.org/abs/2004.05150v2) for more efficient attention computation. **LettuceDetect** capitalizes on ModernBERT’s extended context window to build a token-level classifier for hallucination detection. This approach sidesteps many limitations of older BERT-based models (e.g., short context bounds) and avoids the inference overhead of large LLM-based detectors. Our experiments show that LettuceDetect outperforms other encoder-based systems while remaining **competitive with fine-tuned LLM detectors** at a fraction of their computational cost. ## Data [RAGTruth](https://aclanthology.org/2024.acl-long.585/) is the first large-scale benchmark specifically designed to evaluate **hallucination detection** in Retrieval-Augmented Generation (RAG) settings. It contains **18,000** annotated examples spanning multiple tasks: - **Question Answering (QA)**: Sampled from the [MS MARCO](https://microsoft.github.io/msmarco/) dataset, where up to three documents are retrieved for each question, and an LLM is prompted to produce an answer. - **Data-to-Text Generation**: Based on the [Yelp Open Dataset](https://www.yelp.com/dataset/), where LLMs generate reviews for sampled businesses. - **News Summarization**: Uses random documents from [CNN/DailyMail](https://huggingface.co/datasets/cnn_dailymail), prompting LLMs to provide summaries. Each example in RAGTruth includes: 1. A **question** and **prompt**, 2. One or more **reference passages** or **context documents**, 3. An **LLM-generated response**, 4. **Human annotations** at the span level indicating whether any part of the answer is factually unsupported or contradictory to the reference passages. Below is an illustrative sample from RAGTruth, showing an instance of hallucination: --- **QUESTION** How to explain quartiles? **REFERENCE** > **Passage 2:** > Second quartile (Q2) which is more commonly known as median splits the data in half (50%). > Median divides the data into a lower half and an upper half. > Third quartile (Q3), also known as upper quartile, *splits lowest 75% (or highest 25%)* of data. > *(...)* **RESPONSE** > The first quartile (Q1) splits the lowest 25% of the data, while the second quartile (Q2) splits the data into two equal halves, with the median being the middle value of the lower half. Finally, the third quartile (Q3) splits *the highest 75% of the data.* > *(...)* **ANNOTATION** - **Span**: “highest 75%” - **Type**: Evident Conflict - **Reason**: - Original text: “lowest 75% (or highest 25%)” - Generative text: “highest 75%” --- This example showcases a RAGTruth annotation involving specific spans in the answer that conflict with the reference passages, providing both the **type** of hallucination (e.g., Evident Conflict) and a human-readable **rationale**. Our method only uses binary annotations and does not use the type of hallucination annotations. ## Method

LettuceDetect Architecture

**A high-level depiction of LettuceDetect.** Here, an example **Question**, **Context**, and **Answer** triplet is processed. First, the text is tokenized, after which LettuceDetect performs token-level classification. Tokens from both the question and context are **masked** (indicated by the red line in the figure) to exclude them from the loss function. Each token in the answer receives a probability indicating whether it is hallucinated or supported. For *span-level* detection, we merge consecutive tokens with hallucination probabilities above 0.5 into a single predicted span. --- We train **ModernBERT-base** and **ModernBERT-large** variants as token-classification models on the RAGTruth dataset. The input to the model is a concatenation of **Context**, **Question**, and **Answer** segments, with specialized tokens \([CLS]\) (for the context) and \([SEP]\) (as separators). We limit the sequence length to 4,096 tokens for computational feasibility, though ModernBERT can theoretically handle up to 8,192 tokens. ### Tokenization and Data Processing - **Tokenizer**: We employ _AutoTokenizer_ from the Transformers library to handle subword tokenization, inserting `[CLS]` and `[SEP]` appropriately. - **Labeling**: - Context/question tokens are *masked* (i.e., assigned a label of `-100` in PyTorch) so that they do not contribute to the loss. - Each *answer* token receives a label of **0** (supported) or **1** (hallucinated). ### Model Architecture Our models build on Hugging Face’s _AutoModelForTokenClassification_, using **ModernBERT** as the encoder and a classification head on top. Unlike some previous encoder-based approaches (e.g., ones pre-trained on NLI tasks), our method uses *only* ModernBERT with no additional pretraining stage. ### Training Configuration - **Optimizer**: AdamW, with a learning rate of 1 * 10^-5 and weight decay of 0.01. - **Hardware**: Single NVIDIA A100 GPU. - **Epochs**: 6 total training epochs. - **Batching**: - Batch size of 8, - Data loading with PyTorch _DataLoader_ (shuffling enabled), - Dynamic padding via _DataCollatorForTokenClassification_ to handle variable-length sequences efficiently. During training, we monitor **token-level F1** scores on a validation split, saving checkpoints using the _safetensors_ format. Once training is complete, we upload the best-performing models to Hugging Face for public access. At inference time, the model outputs a probability of hallucination for each token in the answer. We aggregate consecutive tokens exceeding a 0.5 threshold to produce **span-level** predictions, indicating exactly which segments of the answer are likely to be hallucinated. The figure above illustrates this workflow. Next, we provide a more detailed evaluation of the model’s performance. ## Results We evaluate our models on the RAGTruth test set across all task types (Question Answering, Data-to-Text, and Summarization). For each example, RAGTruth includes manually annotated spans indicating hallucinated content. ### Example-Level Results We first assess the **example-level** question: *Does the generated answer contain any hallucination at all?* Our large model (**lettucedetect-large-v1**) attains an **overall F1 score of 79.22%**, surpassing: - **GPT-4** (63.4%), - **Luna** (65.4%) (the previous state of the art encoder-based model), - **Fine-tuned Llama-2-13B** (78.7%) as presented in the RAGTruth paper. It is second only to the fine-tuned **Llama-3-8B** from the **RAG-HAT** paper [15] (83.9%), but LettuceDetect is significantly **smaller** and **faster** to run. Meanwhile, our base model (**lettucedetect-base-v1**) remains highly competitive while using fewer parameters.

Example-level Results

Above is a comparison table illustrating how LettuceDetect aligns against both prompt-based methods (e.g., GPT-4) and alternative encoder-based solutions (e.g., Luna). Overall, **lettucedetect-large-v1** and **lettucedect-base-v1** are very performant models, while being very effective in inference settings. ### Span-Level Results Beyond detecting if an answer contains hallucinations, we also examine LettuceDetect’s ability to identify the **exact spans** of unsupported content. Here, LettuceDetect achieves **state-of-the-art** results among models that have reported span-level performance, substantially outperforming the fine-tuned Llama-2-13B model from the RAGTruth paper [1] and other baselines.

Span-level Results

Most methods, like RAG-HAT [15], do not report span-level metrics, so we do not compare to them here. ### Inference Efficiency Both **lettucedetect-base-v1** and **lettucedetect-large-v1** require fewer parameters than typical LLM-based detectors (e.g., GPT-4 or Llama-3-8B) and can process **30–60 examples per second** on a single NVIDIA A100 GPU. This makes them practical for industrial workloads, real-time user-facing systems, and resource-constrained environments. Overall, these results show that **LettuceDetect** has a good balance: it achieves near state-of-the-art accuracy at a fraction of the size and cost compared to large LLM-based judges, while offering precise, token-level hallucination detection. ## Conclusion We introduced **LettuceDetect**, a lightweight and efficient framework for hallucination detection in RAG systems. By utilizing ModernBERT’s extended context capabilities, our models achieve strong performance on the RAGTruth benchmark while retaining **high inference efficiency**. This work lays the groundwork for future research directions, such as expanding to additional datasets, supporting multiple languages, and exploring more advanced architectures. Even at this stage, LettuceDetect demonstrates that **effective hallucination detection** can be achieved using **lean, purpose-built** encoder-based models. ## Citation If you find this work useful, please cite it as follows: ```bibtex @misc{Kovacs:2025, title={LettuceDetect: A Hallucination Detection Framework for RAG Applications}, author={Ádám Kovács and Gábor Recski}, year={2025}, eprint={2502.17125}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2502.17125}, } ``` Also, if you use our code, please don't forget to give us a star ⭐ on our GitHub repository [here](https://github.com/KRLabsOrg/LettuceDetect). ## References [1] [Niu et al., 2024, RAGTruth: A Dataset for Hallucination Detection in Retrieval-Augmented Generation](https://aclanthology.org/2024.acl-long.585/) [2] [Luna: A Simple and Effective Encoder-Based Model for Hallucination Detection in Retrieval-Augmented Generation](https://aclanthology.org/2025.coling-industry.34/) [3] [ModernBERT: A Modern BERT Model for Long-Context Processing](https://huggingface.co/blog/modernbert) [4] [GPT-4 report](https://arxiv.org/abs/2303.08774) [5] [Llama-3 report](https://arxiv.org/abs/2307.09288) [6] [Mistral 7B](https://arxiv.org/abs/2310.06825) [7] [Kaddour et al., 2023, Challenges and Applications of Large Language Models](https://arxiv.org/pdf/2307.10169) [8] [Huang et al., 2025, A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions](https://dl.acm.org/doi/10.1145/3703155) [9] [Gao et al., 2024, Retrieval-Augmented Generation for Large Language Models: A Survey](https://arxiv.org/abs/2312.10997) [10] [Ji et al., 2023, Survey of Hallucination in Natural Language Generation](https://doi.org/10.1145/3571730) [11] [Sun et al., 2025, ReDeEP: Detecting Hallucination in Retrieval-Augmented Generation via Mechanistic Interpretability](https://arxiv.org/abs/2410.11414) [12] [Manakul et al., 2023, SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models](https://aclanthology.org/2023.emnlp-main.557/) [13] [Cohen et al., 2023, LM vs LM: Detecting Factual Errors via Cross Examination](https://aclanthology.org/2023.emnlp-main.778/) [14] [Friel et al., 2023, Chainpoll: A high efficacy method for LLM hallucination detection](https://arxiv.org/abs/2310.18344) [15] [Song et al., 2024, RAG-HAT: A Hallucination-Aware Tuning Pipeline for {LLM} in Retrieval-Augmented Generation](https://aclanthology.org/2024.emnlp-industry.113/) [16] [Devlin et al., 2019, BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805)