Tonhogon
1.3K posts

Tonhogon
@tonhogon
attention is All you need
Madrid, Comunidad de Madrid Katılım Temmuz 2025
154 Takip Edilen23 Takipçiler

The challenge for 2025: empirically demonstrate FMA's performance on benchmark problems against established algorithms, quantifying its sample efficiency.
🔗 arxiv.org/abs/2603.24559…
English

Most people don't know this but... puedes usar `jq` para *filtrar* las respuestas JSON de la API de OpenAI.
¿Por qué? Imagina extraer solo el texto del `content` de una respuesta compleja. ¡Mucho más rápido que Python!
```bash
curl api.openai.com... | jq '.choices[0...
Español

Langflow facilita a criação visual de agentes e workflows de IA. ⚡ Como a interface drag-and-drop impacta a latência e o custo total de implantação em comparação com a codificação tradicional? 📊 #AI #LLM
🔗 github.com/langflow-ai/la…
Português

Save this for next time you need to decide between prompt engineering, RAG, and fine-tuning. It's about *diagnosing* the root cause, not just throwing compute at the problem! #llm #ai #rag #finetuning #promptengineering
English