carbocation

Biggest market agreed. But relative impact on utility of laptops seems enormous.

Friendly counterpoint (I know, not actually responsive to your comment, but tangentially related): https://www.science.org/doi/10.1126/science.ade4401

This is not a mainstream view of the science, and it's worth noting that this perspective is also not supported by the OP or by the JACC article that it's citing.

Thanks, this is a helpful breakdown.

One thing that is confusing about this write-up is that "DeepConf-low" is only mentioned once and in a screenshot, but it seems to outperform DeepConf-high for several tasks. I guess I'll need to read the underlying paper, but that seems troublesome.

I don't think disclosure is necessary, but I think it can build trust in cases like this. "Please note that we used an LLM to rewrite our initial English draft." The reason to do this is that then people don't waste cycles wondering about the answer to this question.

Actual instance of customer confusion: https://news.ycombinator.com/item?id=44933795

And it's already on ollama, it appears: https://ollama.com/library/gpt-oss

From your first link - I apparently still have 'xnkxglie' memorized...

This does seem like an important omission.

For whatever it's worth, I have found the AI summary of customer reviews helpful.

I think it's neat that this summary is written by an author of the scientific manuscript. Oversimplification is a risk, but this approach eliminates the possibility that the writer did not understand the underlying science.

Methylene blue is a vasopressor, particularly good at reversing vasoplegia after cardiopulmonary bypass.

Is it possible to know which model version their chat app ( https://gemini.google.com/app ) is using?

When I was in my postdoc (applied human genetics), my advisor's rule was that you needed to understand the tools you were using at a layer of abstraction below your interface with them.

For example, if we wanted to conduct an analysis with a new piece of software, it wasn't enough to run the software: we needed to be able to explain the theory behind it (basically, to be able to rewrite the tool).

From that standpoint, I think that even if you keep with #2, you might benefit from taking steps to gain the understanding from #1. It will help you understand the models' real advantages and disadvantages to help you decide how to incorporate them in #2.

All 3 points (two from you and one from '8bitsrule) helped me contextualize this:

* 700 times further from the Sun than the Earth

* 15 times further from the Sun than Pluto

* 0.01 lightyear, or 1/400th the distance to the nearest star

Just following this comment up as a note-to-self: just as `kgwgk noted, the default gemma3:27B model has ID a418f5838eaf, which corresponds to 27b-it-q4_K_M. But the new gemma3:27B quantization-aware training (QAT) model being discussed is gemma3:27b-it-qat with ID 29eb0b9aeda3.

This could actually speed up some of my scientific computing (in some cases, data localization/delocalization is an important part of overall instance run-time). I will be interested to try it.

> The Space Force anticipates awarding 54 launches across the five order years with SpaceX receiving about 60 percent of the missions (28 launches), ULA getting 40 percent (19 launches) and Blue Origin getting seven missions.

Agreed, the content suggest a pretty reasonable spread, especially given each group's historical capabilities.

Thank you. So, another de minimis amount ($1.8 million): it's not exactly zero, but it's just about as much as their NIH support. Columbia, as a comparator, gets $100 million in NSF funding.

I’m not familiar with the NSF funding mechanisms or how people track NSF funding. Not saying NSF is not relevant, just that I’m not using it for my personal heuristic right now.

Wesleyan does not have a medical center and according to the NIH’s public reporting, they have under $2 million in NIH grants, compared to $600 million for Columbia. (Edited from $400 million, which is the value cut.)

Wesleyan has a $250 million operating budget, so the (from what REPORTER indicates) $1.6 million in NIH funding represents 0.6% of their budget. In contrast, the $600 million in NIH funding to Columbia represents about 10% of its $6 billion operating budget.

So both in terms of absolute numbers and relative numbers, the NIH contributions to Wesleyan are de minimis.

So far the fight/not fight decisions can be predicted in advanced based on whether an institution has a medical center with NIH grants.

Thanks, I had not seen that caveat!

This works great for many purposes.

One area where it does not work well at all is modifying photographs of people's faces.* Completely fumbles if you take a selfie and ask it to modify your shirt, for example.

* = unless the people are in the training set

I'm not a python expert, but this feels very odd to me (both the *init* construction and the return [tgemm.mm](http://tgemm.mm/)(input, self.weight, self.bias, None, None) call, which looks like markdown to me:

    from aiter.tuned_gemm import tgemm
    import torch
    
    class LinearLayer(torch.nn.Module):
     def **init**(self, in_features, out_features):
      super(LinearLayer, self).**init**()
      self.weight = torch.nn.Parameter(torch.randn(out_features, in_features).cuda())
      self.bias = torch.nn.Parameter(torch.randn(out_features).cuda())
    
     def forward(self, input):
      input = input.cuda()
      return [tgemm.mm](http://tgemm.mm/)(input, self.weight, self.bias, None, None)

Nova+Serene sounds very metallic at the beginning about 50% of the time for me.

Thanks for posting this. My gut reaction was to say "no, this is a law: the Hatch Act." But the Hatch Act covers partisan political activity while on duty, which is distinct from commercial activity. As you noted, commercial activity is only 'restricted' via a rule.

That's a misunderstanding. This is not a "blog"; this is a press release from a university on its newsfeed. Linking to their own researcher's work is not a matter of being "rigorous" and is unrelated (and not at all similar) to academic citations.

To give you a better understanding of how this works, if my university didn’t include a link to my research in the press release about my publication, I’d contact them to issue a correction. That’s how fundamental it is to link to the research article.