PinnedIntroducing LiqFit: Flexible Few-Shot Learning Library for Cross-Encoder ModelsLiqFit library is a game-changing tool for few-shot learning specifically designed for cross-encoder models. At its core, LiqFit discerns…Jan 30, 2024A response icon2Jan 30, 2024A response icon2
PinnedAchieve 90% Results in Few-Shot Text Classification with Just 0.1% DataZero-shot abilities of modern LLMs are truly inspiring and make us feel that AGI is pretty close. However, it requires large networks and…Dec 27, 2023Dec 27, 2023
PinnedExtract custom table from PDF with LLMsPortable Document Format (PDF) is one of the most widely used file formats for sharing information, especially in academic, scientific…Sep 30, 2023A response icon2Sep 30, 2023A response icon2
GLiNER-Decoder: You extract entities only onceGLiNER is a Named Entity Recognition (NER) model capable of identifying any entity type using a bidirectional transformer encoder…Aug 15, 2025A response icon4Aug 15, 2025A response icon4
Text Classification meets RLClassification is a fundamental ML task aimed at categorizing information, whether text, images, or other data types. While classification…Mar 28, 2025Mar 28, 2025
Pushing zero-shot classification to the limitLast year, we developed a new classification approach inspired by GLiNER, enabling zero-shot classification in a single forward pass —…Mar 7, 2025A response icon1Mar 7, 2025A response icon1
Refreshing zero-shot classification with ModernBERTText classification is one of the fundamental tasks in ML with a long history of research and profound practical interests. Moreover, it’s…Feb 19, 2025A response icon3Feb 19, 2025A response icon3
Replicating DeepSeek R1 for information extractionI have been working on replicating DeepSeek R1 since its release, and here are the early insights I got:Jan 30, 2025Jan 30, 2025
AI Predictions for 2025: A Comprehensive Outlook. Spoiler: Open-Source Will Greatly Win Again.Many of our forecasts for 2024 materialized, making it a remarkable year for AI. We witnessed the emergence of new paradigms such as…Dec 31, 2024Dec 31, 2024
Meet new encoder language models (LLM2Encoder)The evolution of language models has followed two major paths: generative models (such as GPT, Llama and T5) and discriminative models…Sep 10, 2024A response icon2Sep 10, 2024A response icon2