Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Chinchilla asekale Ofin adanwo

Test Chinchilla training-budget findings, the 70B/1.4T example, scaling proportions, approximate compute, and the distinction from deployment optimization.

Jẹmọ itọsọna awọn ọnaAwọn ofin iwọntunwọnsi Chinchilla
Ibeere 1 ti 8

What does Chinchilla's finding that parameters and tokens should scale in similar proportions mean?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. What does Chinchilla's finding that parameters and tokens should scale in similar proportions mean?
    • Both should grow with comparable scaling exponents under the studied training-compute objective
    • The numerical token count must equal the parameter count
    • Training data does not affect the fitted loss
    • Every deployment must choose the largest possible model
  2. What token-per-parameter ratio does the 70B-parameter, 1.4T-token Chinchilla experiment illustrate?
    • Exactly one token per parameter
    • About two thousand tokens per parameter
    • About twenty training tokens per parameter
    • A ratio determined only by the number of GPUs
  3. Awoṣe wo ni Chinchilla ṣe ju bi o ti jẹ pe o kere pupọ?
    • GPT-2
    • Gopher
    • BERT
    • Mamba
  4. Kini Chinchilla tumọ si nipa awọn awoṣe bii GPT-3 ni akoko yẹn?
    • Won ni won overtrained
    • Wọn ni iṣiro kekere ti o wa
    • Wọn ti lo ti ko tọ tokenizer
    • Wọn ko ni ikẹkọ ni ibatan si iwọn wọn
  5. Ni aijọju bawo ni iṣiro ṣe isunmọ ni itupalẹ Chinchilla?
    • Ni ibamu si nọmba awọn GPU nikan
    • Ominira ti data iwọn
    • Iwon to paramita igba ikẹkọ àmi
    • Ni ibamu si iwọn fokabulari
  6. Why might a deployment objective favor a smaller model trained on extra data rather than the training-compute optimum alone?
    • Repeated inference costs can matter alongside the initial training budget
    • Inference has no resource cost for any model
    • The Chinchilla study proved every scaling relationship false
    • More data always reduces both training and serving work to zero
  7. Under the approximation that training compute is proportional to N times D, how must D change if N doubles while that compute budget stays fixed?
    • D must also double
    • D must quadruple
    • D must remain unchanged
    • D must approximately halve
  8. Which three types of term appear in Chinchilla’s parametric loss fit?
    • Only GPU count, electricity price, and elapsed seconds
    • An irreducible base term, a model-size contribution, and a data-size contribution
    • Only vocabulary size, file size, and model name
    • Only a linear token count with no model-size term