GPT: Difference between revisions

75 bytes added ,  2 March 2023
no edit summary
No edit summary
No edit summary
Line 10: Line 10:
[[GPT-3.5]]
[[GPT-3.5]]


==Main GPT Models==
{| class="wikitable"
{| class="wikitable"
|-
|-
Line 15: Line 16:
! rowspan="2" | Parameters
! rowspan="2" | Parameters
! colspan="5" | Training Data
! colspan="5" | Training Data
! Context Window
! colspan="3" | Context Window
|-
|-
! hello
! hello
Line 24: Line 25:
! Tokens
! Tokens
! Words
! Words
! Equivalent
|-
|-
| [[GPT-1]] || 117 Million ||  
| [[GPT-1]] || 117 Million ||  ||  ||  ||  ||  ||
|-
|-
| [[GPT-2]] || 1.5 Billion ||  
| [[GPT-2]] || 1.5 Billion ||  
Line 31: Line 33:
| [[GPT-3]] || 175 Billion ||  
| [[GPT-3]] || 175 Billion ||  
|-
|-
| [[ChatGPT]] ||  ||  
| [[ChatGPT]] ||  || 4,096
|-
|-
| [[GPT-3.5]] ||  ||  
| [[GPT-3.5]] ||  || 4,000
|-
|-
| [[GPT-4]] || ??? ||  
| [[GPT-4]] || ??? ||  
Line 39: Line 41:
|}
|}


==Introduction==
==What is GPT==
[[GPT]], which stands for [[Generative Pre-trained Transformer]], is a type of [[language model]] developed by [[OpenAI]]. Based on the [[Transformer]] [[architecture]] and utilizes [[unsupervised learning]], GPT is able to [[generate text]] indistinguishable from text written by humans.
[[GPT]], which stands for [[Generative Pre-trained Transformer]], is a type of [[language model]] developed by [[OpenAI]]. Based on the [[Transformer]] [[architecture]] and utilizes [[unsupervised learning]], GPT is able to [[generate text]] indistinguishable from text written by humans.