Okay, so there's overwhelming evidence that their public github history is accurate and Nemoclaw was written in a weekend, and the only reason to think it's not accurate is that... it's technically possible to edit git history, and also there's no reasonable explanation for why they would have edited git history they way they did.
So... yeah, draw your own conclusion I guess, whatever.
Lmfao. This is how I know you have never worked at a big company before. I promise you every big company has processes around open sourcing things. It's not something that just whip up and release over a weekend. Just the legal approval would have taken months
I have buddies at Nvidia. Their primary platform is not GitHub. Sorry you're so naive. Almost certainly this was built in house for at least a month or two prior. Then private repo. Approvals. Then public
Not to mention the fact that Jensen literally announced it in their biggest yearly launch conference. No you're totally right. He mandated someone build it over the weekend while drafting up a full presentation and launch announcement about it
That's more plausible than the very normal practice of developing internally, scrubbing commits of any accidental whoopsies, vetting it and then putting it out publicly
"Overwhelming evidence" = git history that is completely fungible. Once you're done here I have a lobster claw to sell you
> Again, I ask you -- what is the reason for them to edit commit history to show incremental progress as if it were written in a weekend, when it actually was not?
Answer this question or we're done here, thanks.
> Almost certainly this was built in house for at least a month or two prior. Then private repo. Approvals. Then public.
Source, other than you making it up?
> That's more plausible than the very normal practice of developing internally, scrubbing commits of any accidental whoopsies, vetting it and then putting it out publicly
Could you point to a specific commit you believe is a representation of an internal data transfer from a separate source control system which is not representative of work achievable within the time period represented by the differential between the commit time and the time of the prior commit?
You cannot really be this naive but i'll play along:
> what is the reason for them to edit commit history to show incremental progress as if it were written in a weekend, when it actually was not?
Like i said. You are letting on that you have never actually worked on an internal project that is going to go open source. There are a million and one reasons. Here are some completely normal and plausible ones. It was worked on over weeks internally, commits referenced other internal NVIDIA software/libraries they used. It name dropped projects and code names. Maybe it was just an extremely long chain of messy commits that is improper to have on a potentially big open source repo. So here's what happens (since you clearly are unaware of how people operate in this world), you "unstage" everything and write canonical commits free of all the garbage. You squash, you merge, you set up standards, you leave a clean commit history. All of it very important for open source
> Source, other than you making it up?
Ah yes let me just go ping the people who worked on it. Lol. Source is my decade long experience working on similar projects where i literally did this scrubbing of commits. You have a circuitous argument "It was done in a weekend because the commits say so" is really quite the hill to die on
> Could you point to a specific commit you believe is a representation of an internal data transfer
If there was any indication left over of a "transfer", it wouldn't have done it's purpose would it? But if you really are looking for something, how about the fact that there's only one human contributor of the first few commits. Very odd, you would think a massive open sourcing of a project like this would probably involve a team right? Or do you believe AI tools have gotten that good that one engineer is just driving with Claude and open sourcing full launches?
Here, how about we just do some critical thinking. Nvidia setup a "Set up NemoClaw" booth at their GTC that was happening just a few days ago. Jensen had a full presentation for it and it was a big highlight.
Do you really think a company as big as Nvidia is hinging the release of a big announcement on the hope that ONE engineer is going to START working on it a few days before the announcement and ACTUALLY get it done to a point where they can talk about it on stage?
Please come on, no one can be this dense. You have to be trolling. Try another argument than "The commits say so". Just apply a basic level of understanding of how software is built and released
> Here are some completely normal and plausible [reasons]. It was worked on over weeks internally, commits referenced other internal NVIDIA software/libraries they used. It name dropped projects and code names. Maybe it was just an extremely long chain of messy commits that is improper to have on a potentially big open source repo.
... it referenced internal servers and they want to scrub that for security reasons
... it might have had secrets embedded at some point because it was a quick and dirty proof-of-concept
... it could have had swear words in the code
... it had enormous binaries checked in at one point and they don't want the repo to be huge
... they don't want you to know the names of everyone that worked on it
... it's forked off other internal work that isn't public yet
There are so many reasons that the easiest thing to do is just snapshot it and have minimal public git history. Some places I've worked make it so publicly, there's one commit per release. Did NVidia do this? Well, they didn't collapse it down to a single commit, but we have no evidence that the commits we see were the actual internal development timeline.
So... yeah, draw your own conclusion I guess, whatever.