jimsimmons

11 ms·

91.
by jimsimmons 3y ago
They didn’t establish a kingdom. And they didn’t affect most of India.
92.
by jimsimmons 3y ago
Muslims didn’t come till 11th century
93.
by jimsimmons 3y ago
Advaita, which my orthodox Brahmin family believes and practices, proclaims the supremacy and totality of the self. Buddhism which was dominant among the elites and aristocrats by 8th century was largely displaced out of India via scholarl
94.
by jimsimmons 3y ago
Sam Altman?
95.
by jimsimmons 3y ago
Definitely. I didn’t mean Americans are toxic. Workplace makes them different people though
96.
by jimsimmons 3y ago
Culture and nationality are different things. Competitiveness and toxicity go hand in hand
97.
by jimsimmons 3y ago
Entire SV VC sphere wants people in the office grinding
98.
by jimsimmons 3y ago
Europeans or literally anyone else are nicer workplace people. American Corp culture is toxic af. Money thirst, shameless self promotion, competition etc really make you anxious all the time.
99.
by jimsimmons 3y ago
Why not. It’s not like they have deadlines
100.
by jimsimmons 3y ago
Exceptions are basically this? Some unknown unknown happened and you can’t do what you wanted to do
101.
by jimsimmons 3y ago
In my college Infosys was always seen as the option only slightly better than unemployment. Actually this was not even the popular view by the time I left.
102.
by jimsimmons 3y ago
Very well articulated.
103.
by jimsimmons 3y ago
Why wouldn’t capitalists drool over obedient, perpetual, cheap labour
104.
by jimsimmons 3y ago
You think Van Gogh and Picasso are nonsense? Or even Mondrian
105.
by jimsimmons 3y ago
He’s talking about the API. Not the web client
106.
by jimsimmons 3y ago
Evernote context seemed great when I used it briefly with my premium subscription. Bubbles up relevant notes automatically without having to link them like in Obsidian or something. I think apart from that or an Anki like system there’s no
107.
by jimsimmons 3y ago
OpenAI dev confirms that API models have not changed: https://twitter.com/OfficialLoganK/status/166393494793189785...
108.
by jimsimmons 3y ago
Well they had every opportunity to inspect its output. Where in the loop and what gets inspected is not an easy problem to solve.
109.
by jimsimmons 3y ago
You have no basis to make that claim. My point was a lot more subtle: if someone asks things like “double check it”, “are you sure” you can provide a template “I’m just a LM” response. I’m not expecting the model to know what it doesn’t kno
110.
by jimsimmons 3y ago
In the loop there indeed was a allegedly trained human in this instance
111.
by jimsimmons 3y ago
They should RLHF this behaviour out. Asking people to be aware of limitations is in similar vein as asking them to read ToC
112.
by jimsimmons 3y ago
Hotel California
113.
by jimsimmons 3y ago
And to not reply in Klingon
114.
by jimsimmons 3y ago
Realistically you can assume there’ll be one soon and develop with these models and switch over
115.
by jimsimmons 3y ago
You provide two samples side by side and see what humans prefer. You should try asking what you don’t know in a non judgemental manner
116.
by jimsimmons 3y ago
The measure is win rate versus DV3. Their model wins more often than ChatGPT Beating a weaker player more often is not evidence of being able to beat a stronger player on average though
117.
by jimsimmons 3y ago
You don’t have 100B params without making a few enemies
118.
by jimsimmons 3y ago
How did this post go from new → top of HN → buried in 3rd page in 1 hour?!
119.
by jimsimmons 3y ago
None of what you say is a gotcha. An ignorant programmer stands only to benefit from Go
120.
by jimsimmons 3y ago
If you’re in denial about that, it’s a you problem

More ›