Adam Kovacs's picture

3 9 23

Adam Kovacs PRO

adaamko

·

AI & ML interests

Explainable AI, rule-based models, hallucination detection, fact-checking LLMs

Recent Activity

upvoted a collection about 1 hour ago

Hallucination detection

posted an update about 1 hour ago

We introduced 🥬 TinyLettuce: lightweight hallucination detection with 17–68M encoders Instead of relying on huge LLM judges that are slow and costly, we built tiny Ettin-based detectors that you can train in hours on a single GPU and run efficiently on CPU. Here’s what’s inside: 1️⃣ Synthetic Data Generator A toolkit to create hallucinations with controllable error types—no manual annotation bottlenecks. 2️⃣ TinyLettuce Models (17–68M) Compact classifiers built on Ettin encoders, designed for efficiency (8K context, modern transformer backbone). 3️⃣ Data & Training Utilities Scripts and APIs to generate domain-specific labeled pairs at scale, plus ~3.6k examples we used for training. 4️⃣ Open and MIT-licensed Code, data, and models are freely available for research and production. 5️⃣ Performance Highlights - TinyLettuce-17M reaches 90.87% F1 (synthetic), outperforming GPT-OSS-120B (83.38%) and Qwen3-235B (79.84%) - Runs in real-time on CPU: low latency, minimal memory, and pennies per million checks - Shows competitive results on RAGTruth benchmarks 🔗 Useful links: Blog: https://huggingface.co/blog/adaamko/tinylettuce GitHub: https://github.com/KRLabsOrg/LettuceDetect PyPI: https://pypi.org/project/lettucedetect/ HF Collection: https://huggingface.co/collections/KRLabsOrg/tinylettuce-68b42a66b8b6aaa4bf287bf4 Notebook/Demo: https://github.com/KRLabsOrg/LettuceDetect/blob/main/demo/tinylettuce.ipynb ⭐ Stars and feedback are always appreciated!

upvoted an article about 2 hours ago

Turk-LettuceDetect: A Hallucination Detection Models for Turkish RAG Applications

View all activity

Organizations

upvoted a collection about 1 hour ago

Hallucination detection

Trained ModernBERT (base and large) for detection hallucinations in LLM responses. The models are trained as token classifications. • 4 items • Updated May 18 • 18

upvoted an article about 2 hours ago

Article

Turk-LettuceDetect: A Hallucination Detection Models for Turkish RAG Applications

By

and 5 others •

4 days ago

• 23

upvoted a collection about 12 hours ago

TinyLettuce

This Collection contains our small, Ettin-encoder (https://arxiv.org/abs/2507.11412) based models trained on synthetic and RagTruth data. • 6 items • Updated 2 days ago • 1

upvoted an article 2 days ago

Article

🥬 TinyLettuce: Efficient Hallucination Detection with 17–68M Encoders

By

and 1 other •

2 days ago

• 5

upvoted a collection 21 days ago

Turkish Hallucination Detection Models

6 items • Updated Jul 23 • 6

upvoted a collection about 1 month ago

Splade Models

The collection includes Splade models from different authors that can be load thanks to the Sparse Encoder modules of Sentence Transformers • 16 items • Updated Jul 30 • 8

upvoted an article 3 months ago

Article

🥬 LettuceDetect Goes Multilingual: Fine-tuning EuroBERT on Synthetic Translations

By

and 1 other •

May 19

• 9

upvoted a collection 4 months ago

Multilingual Hallucination Detection

These are our EuroBERT fine-tunes on our translated RAGTruth datasets. • 13 items • Updated May 18 • 5

upvoted an article 4 months ago

Article

Training and Finetuning Reranker Models with Sentence Transformers v4

By

•

Mar 26

• 160