Government should certainly step back.
Rail and Public transit need a key real estate play.
JR, HK and Singapore prove it.
Even in the early days of the subway in NYC, it drove real estate growth.
Today it's been forgotten or just can't be structured well in the US and many parts of the world, while bureaucrats either run major losses or plop underutilized, poorly integrated stops and systems
I worked on one stage of the JR privatization and this is greatly oversimplifies what actually happened and ignores some important realities.
First, three of the seven JR companies (Hokkaido, Shikoku, Cargo) are still 100% government owned, and were probably never expected to be viable as public enterprises. I was surprised that JR Kyushu was successfully privatized completely - and got Shinkansen - because at the time I was involved (in another company) the conventional wisdom was that only the Honshu companies could succeed as private enterprises.
Second, the privatization also involved parking a large chunk of liabilities in a special entity that was NOT privatized. In fact the part of the privatization I worked on was delayed because the government unilaterally decided the new JR companies had not taken enough of the JNR pension liabilities on their books and added some more. Probably one of the most significant aspects of the privatization was the public service workforce reduction it entailed but that was a problem for the government, not the JR companies.
Third, the JR companies became private (but first government-owned) companies with valuable assets (the commuter and HSR rail networks) build with public funds as well as a portfolio of highly desirable real estate (high traffic train stations) in choice locations.
Fourth, as far as I know any new HSR train lines built since the privatization were publicly funded, since new JR company could afford to build them on a commercially viable basis otherwise.
Fifth, train fares are still regulated, and JR East only recently increased fares for the first time in decades…
Thus, rather than a simplistic public > private transformation, the whole exercise involved a rebalancing of assets, liabilities and risks between the public and private sector, but both remain important to the whole construct. In any case there was serious planning and government decisions around what was necessary to have at least three commercially-viable JR companies (East, West and Tokai) as a result.
So probably more an example of a successfully executed privatization with continuing government involvement, rather than a “private enterprise good, government bad” story.
MedGemma 1.5 is a major upgrade to our open models for healthcare developers.
The new 4B model enables developers to build applications that natively interpret full 3D scans (CTs, MRIs) with high efficiency - a first, we believe, for an open medical generalist model. MedGemma 1.5 also pairs well with MedASR, our speech-to-text model fine-tuned for highly accurate medical dictation.
Developers can now use these multimodal capabilities to build medical apps that reach patients in more places.
Running the full GLM 4.7 (8-bit) on 2 x 512GB M3 Ultra Mac Studios
Runs at 19.8 tok/sec with @exolabs MLX RDMA backend (h/t @awnihannun) & tensor parallel
Running the full GLM 4.7 (8-bit) on 2 x 512GB M3 Ultra Mac Studios
Runs at 19.8 tok/sec with @exolabs MLX RDMA backend (h/t @awnihannun) & tensor parallel
No Pacifica
No HMB
What's the delay/concern?
Poor LIDAR handling of fog?
Economics?
Can be argued that these are reasonable demand areas with many folks looking for a reliable way to get to the city or down peninsula, and currently have no direct BART or convenient transit options
@Waymo@JeffDean
Great use case for @exolabs
Just buy some Mac Mini M4s and link them with some thunderbolt cables (Mac Mini M4s starting at~$450 @microcenter ).
Relatively easy even if you're not very 'techy'.
https://t.co/px0eDUXfsy
📁 Matthew McConaughey says he wants a private LLM, fed only with his books, notes, journals, and aspirations, so he can ask it questions and get answers based solely on that information, without any outside influence.
If you're coding or troubleshooting with @ChatGPTapp, print a PDF of your screen and upload it instead of a screenshot. This preserves your decision tree/prompt edit function, which disappears with image uploads. A PDF parser may use more tokens, but results are better, and retaining prompt editing is useful.