Cuestionario vinculado a una guía · Medio Nivel

Prueba de preentrenamiento continuo versus ajuste fino

Prueba los datos y las diferencias objetivas entre el preentrenamiento continuo y el ajuste, el olvido catastrófico y la repetición, y cuándo la recuperación encaja mejor.

Rutas de guía relacionadasPreentrenamiento continuo versus ajuste fino
Pregunta 1 de 8

¿Cuál es la principal diferencia de datos entre el entrenamiento previo continuo y el ajuste fino supervisado?

Vista previa de cada pregunta8 preguntas

Las respuestas y explicaciones aparecen a medida que realiza el cuestionario.

  1. ¿Cuál es la principal diferencia de datos entre el entrenamiento previo continuo y el ajuste fino supervisado?
    • Both use only labeled question and answer pairs
    • Continued pretraining uses raw domain text; fine-tuning uses curated labeled examples
    • Fine-tuning uses more data than continued pretraining
    • Continued pretraining uses only images
  2. En el ajuste fino supervisado, ¿dónde se calcula normalmente la pérdida?
    • On every token including the prompt
    • Only on the prompt tokens
    • On randomly chosen tokens
    • Only on the response tokens
  3. ¿Qué es la reproducción (replay) en el entrenamiento previo continuo?
    • Mixing some general data into training to limit forgetting
    • Running the same batch twice
    • Restarting training from scratch
    • Replaying user conversations at inference time
  4. ¿Cómo se construyó el código Llama?
    • From scratch on code only
    • By prompt engineering Llama 2
    • By continuing to train Llama 2 on a large code corpus
    • By fine-tuning on a few hundred coding questions
  5. Para hechos que cambian frecuentemente, ¿qué enfoque sugiere la guía que suele ser mejor que el ajuste fino?
    • Continued pretraining every day
    • Retrieval of up-to-date documents
    • Raising the sampling temperature
    • Training a larger tokenizer
  6. Which statement about BloombergGPT is correct according to the guide?
    • It was continued pretraining of Llama 2
    • It was created only with LoRA
    • It was a prompt-engineered GPT-3
    • It was trained from scratch on financial and general data
  7. What did the 2020 paper Don't Stop Pretraining show?
    • A second phase of pretraining on in-domain text improved RoBERTa on domain tasks
    • Pretraining should always stop after one epoch
    • Fine-tuning is useless for classification
    • Domain text harms every model
  8. Why re-warm the learning rate at the start of continued pretraining?
    • To reset the tokenizer
    • To freeze the base weights
    • Resuming at a high rate destabilizes training while too low a rate learns little
    • To skip the need for domain data