Data source: swyx's twitter data
| id | user | created_at | full_text | retweeted_status | quoted_status | place | source | truncated | display_text_range | in_reply_to_status_id | in_reply_to_user_id | in_reply_to_screen_name | geo | coordinates | contributors | is_quote_status | retweet_count | favorite_count | favorited | retweeted | possibly_sensitive | lang | scopes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1431071676918206466 | 33521530 | 2021-08-27T01:50:50+00:00 | @preetster machines are so bad at learning they need a shit ton of data, whereas we regularly learn approximately right lessons from n<10 samples… something i think about often. deep learning is maximum nihilism, next level of ML has to use some form of intuition/universal laws somehow | 95f3aaaddaa45937ac94765e0ddb68ba2be92d20 | 0 | [11, 289] | 1431063186157887493 | 16343912 | preetster | 0 | 0 | 1 | 0 | 0 | en | ||||||||
| 1431074678613807107 | 33521530 | 2021-08-27T02:02:46+00:00 | @preetster yeah like yes transfer learning is a thing but it feels positively neanderthal compared to what human babies can do | 95f3aaaddaa45937ac94765e0ddb68ba2be92d20 | 0 | [11, 126] | 1431072993774505986 | 16343912 | preetster | 0 | 0 | 0 | 0 | 0 | en |