Dylan

[Orson Scott Card @orsonscottcard]

The stories in ENDER’S TRIBE were published in various magazines or anthologies in the past but are now hard to find. Everything else I’ve written in the “Enderverse” is already in book form, so it was time for these to be collected into one too (thanks to Blackstone Publishing @BlackstoneAudio).
The original Ender’s Game short story has been reprinted most, but it seemed right to include it here as well.

[Dylan @DylanBForde]

@HSVSphere @shaurizard Why are people saying this? With current AV you could likely just pop the car into the air to avoid an oncoming collision. I would always take a single use if the single use is an extra life tbh

[Dylan @DylanBForde]

What is bernardo silva doing here

[Dylan @DylanBForde]

@MaxDemian36 @nntaleb It probably is a genuine concern for people in the industry, which is genuinely a bit crazy incentives are so mismatched

[Dylan @DylanBForde]

@AndyMasley I could understand the point of view for not watermarking text, images otohshould be watermarked imo. However for protein synethesis I would say it's the most important modality to watermark.

[Dylan @DylanBForde]

@prof_kamilov My first ever conference submission, single author neurips, was accepted and I feel like I have pre war steel.

[Dylan @DylanBForde]

@Bryan_1269 @bubbleboi Yeah originally I did an MVP recently in colab and then moved it over to a gcp instance with 8 tpu after. Lost about 100 when I found out the sharding wasn't working and stopped it on the hour. The ring topology they have is quite annoying vs a dense connection. Otherwise, tpu >

[Dylan @DylanBForde]

@Bryan_1269 @bubbleboi It's definitely improved substantially from when I first tried it, much more forgiving but generally still cutting edge and experimental leading to some issues. Generally ok though.

[Dylan @DylanBForde]

NeurIPS this year staying around 25% of submissions accepted.

[Dylan @DylanBForde]

@Bryan_1269 @bubbleboi Have you tried Pallas ? Few bugs but mostly decent

[閈姬 @SandersIrvingP1]

@abyandaniel @at_fome 不用担心，这个领域大量伪造的论文会对LLM造成显著的投毒效果😁

[Dylan @DylanBForde]

@cHHillee All are just proxies for scale though

[Dylan @DylanBForde]

Very maximum upper bound imo, this is assuming everyone uses the allowance and that 100% of available compute is for subscribers and none is for training/research.

[Dylan @DylanBForde]

People saying this used a lot of compute, I hardly understand how you can think that.

They used 88hrs • 10000agents (880000 hrs of agents). They have approximately 50M subscribers (rumours). You get about 10hrs per week in the 20pm of Astra ultra.

This is about .18% compute

[OpenAI @OpenAI]

We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.

The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.

The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.

[Dylan @DylanBForde]

Potential revenue at max is like 4M and cost is 1M (said) so missing out on only 3M profit to solve this problem

[Dylan @DylanBForde]

Nice

https://t.co/vzAfg4Wqyk

[Dylan @DylanBForde]

@hellothxok @ArmedJ0y Thanks, the only people I know who carried had the same opinion. Nice to know it was legitimate

[Dylan @DylanBForde]

Maybe it's too lighthearted but they should have given PHASEONE[big] the report and asked it to apologize. Tragic that it's own opinion is not included.

[Dylan @DylanBForde]

@dadkins_ Poor guy got spawned into crime