urchade (UZ)

GLiNER has now 400+ stars on GitHub (github.com/urchade/GLiNER) #NLP #NER

Fine-tune SLMs for agents for free this Friday. You don’t even need to bring a dataset, just show up, write a few prompts and walk out with a fine-tuned SLM. @AWS, @OpenAI, @Render, @Modulate, and @Neo4j will also be there. $47k+ in prizes. See you there.

Next week I'm speaking alongside Sid Bidasaria (co-creator of Claude Code) and @hwchase17 from @LangChain at Coding Agents by @mlopscommunity My talk: why agents need small language models. 📅 March 3rd 📍 Computer History Museum, Mountain View 🔗 luma.com/codingagents

DC what's up!

"Stop me when I say a better shooter." "Steph Curry?" Tre Johnson: "Nah, he knows I'm better... I'm just gonna show him." 👀 (via @MonSportsNet / YouTube)

🥱

@DjidjeliF Done ✅

👋 Bonjour @urchadeDS Je viens d’ouvrir une PR pour ajouter la version safetensors du modèle GLiNER (gliner_multi-v2.1) 👉 huggingface.co/urchade/gliner… Ce format est plus sûr et recommandé pour l’inférence. Peux-tu la valider lorsque tu auras un moment ? 🙏😊

Unifies entity recognition, classification, and extraction github.com/fastino-ai/GLi…

In GLiNER2, you can combine multiple tasks in a single model call 🐙. In this example, I'm parsing a hotel booking query and simultaneously extracting entities (hotel name, dates...) 📍 while performing text classification. 💡 Try it: github.com/fastino-ai/GLi…

Early preview from from our researcher @var6595 of a new foundational model for personalization we've been building at Fastino. At the self-evolving agents hack luma.com/agentshack

Introducing GLiNER-2 - Fastino’s next-gen open-source model for unified entity extraction, classification & structured parsing. • NER, classification & JSON in 1 blazing-fast pass • ⚡ <150 ms CPU latency • 🧩 Apache-2.0 + hosted API Built by @fastino_ai — unveiled live at #EMNLP2025 by @urchadeDS 🔗 github.com/fastino-ai/GLi…

Big milestone for our team - GLiNER-2 is live! One model for NER, classification & structured parsing in a single pass. <150 ms CPU latency, open-source (Apache-2.0) + hosted API. 🔗 github.com/fastino-ai/GLi… Incredible work by @urchadeDS and the @fastinoAI research crew after EMNLP 2025 🙌

This paper builds a tough test for proactive agents and shows current models still fail often. The test is called PROBE (Proactive Resolution of Bottlenecks) and it checks 3 steps, search, identify the real problem, execute the fix. Each task gives a big pile of emails, docs, and calendar items with 1 hidden bottleneck. A model must find the right evidence, name the exact blocker and task, then pick 1 action with full parameters. The dataset has 1000 cases with long, noisy context so shortcuts break. Top models reach about 40% on full end to end success. Agent frameworks that chain tools do worse because retrieval misses key evidence. Most errors come from wrong root cause, wrong person, missed deadlines, or missing action parameters. The main weakness is turning messy evidence into a precise, complete action. So proactive help is far from solved in real work like settings. ---- Paper – arxiv. org/abs/2510.19771 Paper Title: "Beyond Reactivity: Measuring Proactive Problem Solving in LLM Agents"

New state-of-the-art Personally Identifiable Information extraction that can run extremely efficiently, even on CPUs. In my opinion, PII is one of the most important entity recognition tasks, and the GLiNER architecture is absolute best option for it right now.

🚀 Our largest study on zero-shot text classification is out! 📄 arxiv.org/pdf/2508.07662 We surpass cross-encoders while being much faster, especially for large label sets. Check out all the research results 👇

😎 I just published Sentence Transformers v5.1.0, and it's a big one. 2x-3x speedups of SparseEncoder models via ONNX and/or OpenVINO backends, easier distillation data preparation with hard negatives mining, and more! See 🧵for the deets:

code soon 👀

@ash_csx Huge thanks, boss! Really proud to be part of Fastino 🦊 You and @george_onx were incredibly supportive during the thesis writing and defense, truly grateful!

Huge congrats to @urchadeDS on winning the 2025 ATALA Thesis Prize (best NLP thesis in France)! An honor to work with you 🦊

Locked in and having fun with the team 🦊😤🔐

Starting the 2025 @fastinoAI team summit! 🇬🇧🦊