GPT-4: Difference between revisions
Maxeto0910 (talk | contribs) mNo edit summary Tags: Visual edit Mobile edit Mobile web edit Advanced mobile edit |
Encourage secondary sources |
||
Line 3: | Line 3: | ||
{{Infobox software |
{{Infobox software |
||
| title = Generative Pre-trained Transformer 4 (GPT-4) |
| title = Generative Pre-trained Transformer 4 (GPT-4) |
||
| author = [[OpenAI]] |
| author = [[OpenAI]] |
||
| released = March 14, 2023 |
| released = March 14, 2023 |
||
| genre = [[Autoregressive]] [[Transformer (machine learning model)|transformer]] [[language model]] |
| genre = [[Autoregressive]] [[Transformer (machine learning model)|transformer]] [[language model]] |
||
}} |
}} |
||
'''GPT-4''' ('''Generative Pre-trained Transformer 4''') is a [[multimodal learning|multimodal]] [[large language model]] created by [[OpenAI]],<ref name=" |
'''GPT-4''' ('''Generative Pre-trained Transformer 4''') is a [[multimodal learning|multimodal]] [[large language model]] created by [[OpenAI]],<ref name="ars-technica">{{Cite web |last=Edwards |first=Benj |date= March 14, 2023 |title=OpenAI’s GPT-4 exhibits “human-level performance” on professional benchmarks |url=https://arstechnica.com/information-technology/2023/03/openai-announces-gpt-4-its-next-generation-ai-language-model/ |access-date=March 15, 2023 |website=[[Ars Technica]] }}</ref> the fourth in the [[Generative pre-trained transformer|GPT]] series. It was released on March 14, 2023, and will be available via [[API]] and for [[ChatGPT]] Plus users.<ref name="ars-technica" /> [[Microsoft]] confirmed that versions of [[Microsoft Bing|Bing]] using GPT had in fact been using GPT-4 before its official release.<ref>{{Cite web |date=March 14, 2023 |url=https://techcrunch.com/2023/03/14/microsofts-new-bing-was-using-gpt-4-all-along/ |title=Microsoft’s new Bing was using GPT-4 all along |last=Lardinois |first=Frederic |website=TechCrunch |access-date=March 14, 2023 |archive-date=March 15, 2023 |archive-url=https://web.archive.org/web/20230315013650/https://techcrunch.com/2023/03/14/microsofts-new-bing-was-using-gpt-4-all-along/ |url-status=live }}</ref> As a [[Transformer (machine learning model)|transformer]], GPT-4 was pretrained to predict the next token (using both public data and "data licensed from third-party providers"), and was then fine-tuned with [[reinforcement learning from human feedback]].<ref name="GPT4Tech" />{{Rp|page=2}} |
||
The GPT-4 technical report explicitly refrained from specifying the model size, citing "the competitive landscape and the safety implications of large-scale models".<ref name="GPT4Tech" /> ''[[The Verge]]'' cited rumors that GPT-4 would substantially increase the parameter count from [[GPT-3]]'s 175 billion to 100 trillion, which [[OpenAI]] CEO [[Sam Altman]] described as "complete bullshit".<ref name="verge" /> U.S. Representatives [[Don Beyer]] and [[Ted Lieu]] confirmed to the [[New York Times]] that Altman visited [[United States Congress|Congress]] in January 2023 to demonstrate GPT-4 and its improved "security controls" compared to other AI models.<ref name="nyt-3" /> |
The GPT-4 technical report explicitly refrained from specifying the model size, citing "the competitive landscape and the safety implications of large-scale models".<ref name="GPT4Tech" /> ''[[The Verge]]'' cited rumors that GPT-4 would substantially increase the parameter count from [[GPT-3]]'s 175 billion to 100 trillion, which [[OpenAI]] CEO [[Sam Altman]] described as "complete bullshit".<ref name="verge" /> U.S. Representatives [[Don Beyer]] and [[Ted Lieu]] confirmed to the [[New York Times]] that Altman visited [[United States Congress|Congress]] in January 2023 to demonstrate GPT-4 and its improved "security controls" compared to other AI models.<ref name="nyt-3" /> |
||
[[OpenAI]] wrote in their blog post announcing GPT-4 that "GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than [[ChatGPT|GPT-3.5]]."<ref |
[[OpenAI]] wrote in their blog post announcing GPT-4 that "GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than [[ChatGPT|GPT-3.5]]."<ref>{{Cite web |last=Wiggers |first=Kyle |date=March 14, 2023 |title=OpenAI releases GPT-4, a multimodal AI that it claims is state-of-the-art |url=https://techcrunch.com/2023/03/14/openai-releases-gpt-4-ai-that-it-claims-is-state-of-the-art/ |access-date=March 15, 2023 |website=TechCrunch }}</ref> It can read, analyze, or generate up to 25,000 words of text, which is a significant improvement over previous versions of the technology.<ref name="ars-technica" /><ref>{{cite AV media|title=GPT-4 Developer Livestream|date=March 14, 2023|type=Videotape|publisher=OpenAI|via=YouTube|url=https://www.youtube.com/watch?v=outcGtbnMuQ&ab_channel=OpenAI|access-date=14 March 2023|language=en|archive-date=March 14, 2023|archive-url=https://web.archive.org/web/20230314172139/https://www.youtube.com/watch?v=outcGtbnMuQ&ab_channel=OpenAI|url-status=live}}</ref> The ''New York Times'' wrote that GPT-4 showed impressive improvements in accuracy compared to GPT-3.5, had gained the ability to summarize and comment on images, was able to summarize complicated texts, passed a [[Bar examination|bar exam]] and several standardized tests, but still showed a tendency to [[hallucination (artificial intelligence)|hallucinate]] answers.<ref name="nyt-4">{{Cite news |last=Metz |first=Cade |last2=Collins |first2=Keith |date=March 14, 2023 |title=10 Ways GPT-4 Is Impressive but Still Flawed |language=en-US |work=The New York Times |url=https://www.nytimes.com/2023/03/14/technology/openai-new-gpt4.html |access-date=March 14, 2023 |archive-url=https://archive.is/fL9jH |archive-date=March 14, 2023 |issn=}}</ref> |
||
== References == |
== References == |
||
Line 40: | Line 40: | ||
| url-status = live |
| url-status = live |
||
}}</ref> |
}}</ref> |
||
⚫ | <ref name="GPT4Tech">{{Cite web |date=2023 |title=GPT-4 Technical Report |url=https://cdn.openai.com/papers/gpt-4.pdf |website=[[OpenAI]] |access-date=March 14, 2023 |archive-date=March 14, 2023 |archive-url=https://web.archive.org/web/20230314190904/https://cdn.openai.com/papers/gpt-4.pdf |url-status=live }}</ref> |
||
<ref name="product-page">{{Cite web |
|||
|date=March 14, 2023 |
|||
|title=GPT-4 |
|||
|url=https://openai.com/product/gpt-4 |
|||
|url-status=live |
|||
|access-date=March 14, 2023 |
|||
|website=[[OpenAI]] |
|||
|archive-date=March 14, 2023 |
|||
|archive-url=https://web.archive.org/web/20230314165712/https://openai.com/product/gpt-4 |
|||
}}</ref> |
|||
⚫ | <ref name="GPT4Tech">{{Cite web |date=2023 |title=GPT-4 Technical Report |url=https://cdn.openai.com/papers/gpt-4.pdf |website=[[OpenAI]] |access-date= |
||
}} |
}} |
||
{{Differentiable computing}} |
{{Differentiable computing}} |
Revision as of 11:00, 15 March 2023
Original author(s) | OpenAI |
---|---|
Initial release | March 14, 2023 |
Type | Autoregressive transformer language model |
Website | openai![]() |
GPT-4 (Generative Pre-trained Transformer 4) is a multimodal large language model created by OpenAI,[1] the fourth in the GPT series. It was released on March 14, 2023, and will be available via API and for ChatGPT Plus users.[1] Microsoft confirmed that versions of Bing using GPT had in fact been using GPT-4 before its official release.[2] As a transformer, GPT-4 was pretrained to predict the next token (using both public data and "data licensed from third-party providers"), and was then fine-tuned with reinforcement learning from human feedback.[3]: 2
The GPT-4 technical report explicitly refrained from specifying the model size, citing "the competitive landscape and the safety implications of large-scale models".[3] The Verge cited rumors that GPT-4 would substantially increase the parameter count from GPT-3's 175 billion to 100 trillion, which OpenAI CEO Sam Altman described as "complete bullshit".[4] U.S. Representatives Don Beyer and Ted Lieu confirmed to the New York Times that Altman visited Congress in January 2023 to demonstrate GPT-4 and its improved "security controls" compared to other AI models.[5]
OpenAI wrote in their blog post announcing GPT-4 that "GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than GPT-3.5."[6] It can read, analyze, or generate up to 25,000 words of text, which is a significant improvement over previous versions of the technology.[1][7] The New York Times wrote that GPT-4 showed impressive improvements in accuracy compared to GPT-3.5, had gained the ability to summarize and comment on images, was able to summarize complicated texts, passed a bar exam and several standardized tests, but still showed a tendency to hallucinate answers.[8]
References
- ^ a b c Edwards, Benj (March 14, 2023). "OpenAI's GPT-4 exhibits "human-level performance" on professional benchmarks". Ars Technica. Retrieved March 15, 2023.
- ^ Lardinois, Frederic (March 14, 2023). "Microsoft's new Bing was using GPT-4 all along". TechCrunch. Archived from the original on March 15, 2023. Retrieved March 14, 2023.
- ^ a b "GPT-4 Technical Report" (PDF). OpenAI. 2023. Archived (PDF) from the original on March 14, 2023. Retrieved March 14, 2023.
- ^ Vincent, James (January 18, 2023). "OpenAI CEO Sam Altman on GPT-4: "people are begging to be disappointed and they will be"". The Verge. Archived from the original on January 26, 2023. Retrieved January 27, 2023.
- ^ Kang, Cecilia (March 3, 2023). "As A.I. Booms, Lawmakers Struggle to Understand the Technology". The New York Times. Archived from the original on March 3, 2023. Retrieved March 3, 2023.
- ^ Wiggers, Kyle (March 14, 2023). "OpenAI releases GPT-4, a multimodal AI that it claims is state-of-the-art". TechCrunch. Retrieved March 15, 2023.
- ^ GPT-4 Developer Livestream (Videotape). OpenAI. March 14, 2023. Archived from the original on March 14, 2023. Retrieved 14 March 2023 – via YouTube.
- ^ Metz, Cade; Collins, Keith (March 14, 2023). "10 Ways GPT-4 Is Impressive but Still Flawed". The New York Times. Archived from the original on March 14, 2023. Retrieved March 14, 2023.