DataChazGPT 🤯 (not a bot)
Oh, the things you’ll do with the new #ChatGPT-powered @Bing! 🪄✨
✔️ The search bar's now a chatbox
✔️ Bing does research for you
✔️ Bing can make plans
✔️ Ask Bing to be creative
✔️ Interact with Bing like a chatbot
✔️ Or... search the traditional way!
https://t.co/yxvuRrUT0G
tweet
Oh, the things you’ll do with the new #ChatGPT-powered @Bing! 🪄✨
✔️ The search bar's now a chatbox
✔️ Bing does research for you
✔️ Bing can make plans
✔️ Ask Bing to be creative
✔️ Interact with Bing like a chatbot
✔️ Or... search the traditional way!
https://t.co/yxvuRrUT0G
tweet
Offshore
Photo
DataChazGPT 🤯 (not a bot)
RT @DataChaz: 'more persuasive', 'concise' or 'action-oriented', here are 7 key phrases that can help you get better rewrites from #ChatGPT! 🤯👇
- prompt idea by @dickiebush https://t.co/efNAiYzhCi
tweet
RT @DataChaz: 'more persuasive', 'concise' or 'action-oriented', here are 7 key phrases that can help you get better rewrites from #ChatGPT! 🤯👇
- prompt idea by @dickiebush https://t.co/efNAiYzhCi
tweet
Offshore
Photo
Aran Komatsuzaki
Has anyone tried next token prediction with truncated BPTT + label prediction at the end for a Long Range Arena like task?
I think TBPTT setting is interesting because some real-world tasks may have longer context than the training-time data.
Ref: https://t.co/IsKHLu9mSo https://t.co/sYmb2bCJc6
tweet
Has anyone tried next token prediction with truncated BPTT + label prediction at the end for a Long Range Arena like task?
I think TBPTT setting is interesting because some real-world tasks may have longer context than the training-time data.
Ref: https://t.co/IsKHLu9mSo https://t.co/sYmb2bCJc6
tweet