Autoregressive Generation
التوليد الارتجاعي
أسلوب توليد يُنتج فيه النموذج رمزاً واحداً في كل خطوة مستنداً إلى جميع الرموز السابقة، مكرراً العملية حتى اكتمال التسلسل.
A generation method where the model produces one token at a time conditioned on all previous tokens, repeating until the sequence is complete.
Also translated asالتوليد التلقائي الارتجاعي، التوليد التسلسلي
First appears in this corpus in: Evaluating Large Language Models Trained on Code (2021)
Appears in these papers
- Evaluating Large Language Models Trained on Code2021in the sky ✦
- GPTQ: Accurate Post-Training Quantization for Generative Pre-Trained Transformers2022in the sky ✦
- Llama 2: Open Foundation and Fine-Tuned Chat Models2023in the sky ✦
- Mistral 7B2023in the sky ✦
- Efficient Memory Management for Large Language Model Serving with PagedAttention2023in the sky ✦
- PaLM: Scaling Language Modeling with Pathways2022in the sky ✦
- RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control2023in the sky ✦
- Sparks of Artificial General Intelligence: Early Experiments with GPT-42023in the sky ✦
- Fast Inference from Transformers via Speculative Decoding2023in the sky ✦
- Fast Inference from Transformers via Speculative Decoding2023in the sky ✦