Calibration
المعايرة
مدى مطابقة ثقة النموذج المتوقعة لدقته الفعلية: النموذج جيد المعايرة الذي يقول «واثق 80%» يكون محقاً 80% من الوقت فعلاً.
How well a model's predicted confidence matches its actual accuracy: a well-calibrated model that says '80% sure' should be correct 80% of the time.
Also translated asتقييس الاحتمالات المسندة، ضبط الثقة، ضبط وثوقية المخرجات العددية للنموذج، معايرة النموذج
First appears in this corpus in: When Does Label Smoothing Help? (2019)
Appears in these papers
- GPT-4 Technical Report2023in the sky ✦
- GPT-4 Technical Report2023in the sky ✦
- GPTQ: Accurate Post-Training Quantization for Generative Pre-Trained Transformers2022in the sky ✦
- When Does Label Smoothing Help?2019in the sky ✦
- When Does Label Smoothing Help?2019in the sky ✦
- LLM.int8(): 8-Bit Matrix Multiplication for Transformers at Scale2022in the sky ✦
- Self-Consistency Improves Chain of Thought Reasoning in Language Models2022in the sky ✦
- Sparks of Artificial General Intelligence: Early Experiments with GPT-42023in the sky ✦
- Sparks of Artificial General Intelligence: Early Experiments with GPT-42023in the sky ✦
- WebArena: A Realistic Web Environment for Building Autonomous Agents2023in the sky ✦