Lann Yecun

[Lann Yecun @LYecun]

Almost all the methods that NLP people cautioned against with claims like ‘You can’t X’ have proven to be possible, just hard. ‘LLMs can’t judge’, ‘Don’t train models on synthetic data’ and ‘Don’t make big models.’ All these methods are now standard practices.

[Lann Yecun @LYecun]

@ylecun Seems fair to me.