Absolutely unacceptable that their TOS basically gives them unrestricted access to your datasets, as far as I read it. The terms let them use your datasets for pretty much whatever they decide they want to do with it (though they do say they would anonymize the data, which isn't especially helpful). The TOS leaves a lot of wiggle room for them to do pretty much whatever they want to the data.
I wouldn't touch this until they get serious about having real assurances that they're not going to access customer data without a real, justifiable reason. If Amazon gave themselves free reign to read S3 data it would be outrageous, this is basically the same thing.
Your datasets are some of your most valuable assets. How is it entitled to not want your vendor having a free for all with them? It would be entitled if they weren't going to be charging for it.
My research group at Stanford has been alpha testing Tinker, it's both very useful and also really technically impressive in my opinion. It's a unified framework for post-training models and it abstracts almost all of the complexity of managing these jobs across resources. That it manages to do this while also allowing a lot of algorithmic flexibility is pretty unique.
Interesting that their first product is an infrastructure play. Is it really so hard to set up a fine-tuning pipeline for yourself that a $12 billion startup with unlimited hype needs to be offering it? Maybe they have figured, whether correctly or not, that building AI tooling is going to be more lucrative than the AI itself.
The thing about this that’s interesting to me is that it can be used as a foundation for products they or other people make that combine real time RL rewards and fine tuning to improve the model. I see a lot of potential here compared to the standard paradigm of ChatGPT wrappers that involve tweaking the prompt or harness to improve it, which is a lot more constrained.
"the Tinker API provides simple functions to compute gradients, update the weights, and sample outputs from the trained model"
It sure sounds like a PyTorch tutorial, but I believe it's yet another "AI training made slightly easier for you" start-up. But all of them seem to solve the easy problem of managing data and compute, while very few tackle the hard problem of generating good training data.
Commonly attempting the "private beta with a waitlist" pseudo-release model (until they finally learned their lesson relatively recently) is a large part of how google fumbled the LLM ball to OpenAI and others.
This is great as we need more solutions that help people train and fine-tune models. There is a lot of open source and some companies behind some of the popular open source packages, like Unsloth, but the more the merrier, esp given that Thinking Machines has the expertise and resources to build something that last.
Yeah, I'm really curious about their stacked multi-tenant lora training at the same time. If this gets commoditised enough, it could be interesting to try "end of the day fine-tunes on daily conversations" and see where that leads. Or a targeted RL on "missed / rejected tasks" for an agent, after you get enough samples for a run, and so on.
1. Can someone help me articulate what Tinker can do that Vertex AI or many others can't? (I can see access to some primitives, which is nice)
2. and more broadly: Has anyone got real lift in business metrics through fine-tuning an open model over using the flagship models from say OpenAI or Anthropic?
Most managed finetuning offerings take your dataset, some hyperparameters, and spit out a model. Few support RL, and those that do have very limited support.
And I have gotten a real lift, in cost effectiveness and engagement (for creative writing)
I don't apply RL directly to engagement (and don't think it's really possible without some insane scale of feedback)
Instead there are mechanical mistakes models make that harm engagement and are trivially verifiable (overused phrases and concepts, hitting a given target reading level, etc.)
Funny timing that they are announcing their first product days after Matt Levine highlighted their lack of a public product or direction in the Money Stuff newsletter.
It's the first that came to mind when I saw the domain name but I'm sure he's not the first to point it out. So it probably is just inevitable that they'd be launching after someone has mentioned it during the last couple of hours/days.
There was a famous tech company in supercomputing and AI, called "Thinking Machines", worked at by people such as Danny Hillis and even Richard Feynman.
Does this new company have some connection to that, such as some of the same people?
> Does this new company have some connection to that, such as some of the same people?
Really doubt it. This "Thinking Machines" seems like an on-trend SV AI software startup, that "Thinking Machines" was a Massachusetts-based hardware company that went bankrupt 30 years ago (and whatever's left is Oracle). That "Thinking Machines" people are retiring by now.
I wonder if Sun/Oracle let the trademark lapse, because if it's still active I'd imagine this startup's gonna get sued.
nope, this is Mira Murati's thinking machines. she used to be the CTO of OpenAI, but started her own thing once she [I think] realized that she was missing out on the gold rush by staying there.
That's a fair valuation if they have a >10% chance of developing a frontier model. It's an excessive valuation if they just release tools to create LoRAs and such.
Usually we want an announcement to come with more than a waitlist before doing a frontpage thread on HN, but I guess this company is high-profile enough that the post is relevant anyway?
I’ve never heard of this in my life. Isn’t tinker a verb meaning to fiddle or edit or modify in small increments? That seems like the perfect name given what the software they’re present. In any case I guess this is back to the debate: does it matter how a word is intended or does it matter how a word is received?
Back in the '90s, people would say "oh he's 'special'" as a slur.
That being said, in a world like we live in today, pretty much anything you say or do will offend someone for some obscure reason that you just can't reasonably anticipate.
It's in these contexts that I think the most appropriate response is "get a life."
> does it matter how a word is intended or does it matter how a word is received?
If you apply that argument to an ethnic slur that's common where you live, you'll see that it wouldn't be a good product name in an international market.
> The term "tinker" is a racial slur when used against Irish and Scottish Travellers and Romani people. Originally derived from the name of an itinerant profession, the word evolved into a derogatory ethnic insult with connotations of being dirty, dishonest, and criminal.
I'm an American of Irish and Scottish descent with some travellers in my background. Never heard this word as a slur, only as "I like to tinker with machines." Tinker as a slur hasn't travelled off the islands of Great Britain apparently.
Edit: I'm not changing my usage of the word. I like to tinker.
I'm very aware of how I speak which is why I have a good track record of not offending people. I change with the times e.g. I no longer refer to a car transmission as a tranny. So I know tinker as a slur is archaic. I've never heard anyone use it as a slur and think it would be counter productive to revive it as one.
You not hearing it doesn’t make it archaic. Tinker is still thrown at Travellers in Ireland, the UK, Australia, and elsewhere. Calling it “dead” just shows you your bubble, not reality.
I think you're now being deliberately obtuse or trolling, if you can't understand the difference between naming a product an ethnic slur and the colloquial use of a swear word.
On the contrary, it seems more acceptable now than 37 years ago - the “unacceptable” meaning that was already residual then (and not mentioned in the article at all!) is even more irrelevant today.
I think it's used less than 37 years ago in that sense - and it was not much of an issue back then (at least in the US, that article doesn't mention that tinker could be considered offensive while it talks about 'wetbacks').
You believe the slur is used now more than 37 years ago or something?
Do you have a source for this? I can't find any etymology dictionary that says it doesn't come from either "tin" as in the metal or "tink" as an onomatopoeia (or a verb that refers to mending things). To be fair, they say it's uncertain, but you seem very confident about your alternative etymology.
The formation etymology (whether from tin or onomatopoeia) is uncertain. The part that is certain is the semantic chronology. The noun tinker was used from at least the 13th century for an itinerant mender of pots, the Travellers. By the 16th century it became a slur for Travellers.
The verb to tinker doesn’t appear until the mid-17th century, first meaning to work as a tinker and only later coming to mean what you're familiar with.
So while the root word’s sound-shape is debated, the order of senses is clear: the Traveller sense comes first, the modern “casual repair” sense comes later and was derived from it. This is the etymological order given in all sources, eg https://www.merriam-webster.com/dictionary/tinker
Does this page show something different in your region? For me it doesn't say anything about what you're claiming it does outside of one "chiefly Ireland, sometimes offensive" definition. The etymology only says "Middle English tinkere", and the history explicitly states its first use as being the not-"chiefly Ireland, sometimes offensive" definition. The etymologies I was seeing show it going from "this is a word that describes a job" and branching to "this group of people does this job a lot, let's call them this word" and "fiddling with things to do anything is close enough".
I'm genuinely interested in this, I work in what's a relatively "woke" domain (education) and I've never heard a complaint about something being called "tinkerable", even from colleagues in the UK.
Let's say you're correct. If "the etymological order given in all sources" is yours, shouldn't you be able to provide an example of that instead of one that requires assuming you're correct and reading words that aren't in the definition, while ignoring the listed etymology?
> the order of senses is clear: the Traveller sense comes first, the modern “casual repair” sense comes later and was derived from it.
The order of the senses is clear but different:
The "mender of kettles, pots, pans, etc.," sense come first.
The “gipsy” sense comes later an is derived from it.
The “repair or put into shape rudely or temporarily" - and later "work imperfectly, work in an experimental or meddlesome manner; keep busy in a useless way" - sense also comes from the first one.
[…] is to be considered that genuine Gypsies have often been spoken of as "tinkers" (chaudronniers) on account of the occupation with which they have long been associated ; and that, although there is no known mention of "Gypsies" in the British Islands prior to the fifteenth century, there are many earlier references to "tinkers" or "tinklers," as they are called in Scotland. […] "Tinkler can be traced back to about the year 1200. Tinker and Tinkler were not uncommon titles at that time. […] All these seem to have had fixed abodes, and not to have been of the same itinerant class with which we now associate all tinkers, and which used to require the epithet 'wandering' to distinguish them." […] To the same purpose as the opinion expressed in this last sentence is Crofton's observation made elsewhere, that "all Gypsies may be pedlars, brasiers, or tinkers, but the reverse may not follow."
[ Scottish Gypsies under the Stewarts; David MacRitchie; 1894 ]
> 1) The "mender of kettles, pots, pans, etc.," sense come first.
> 2) The “gipsy” sense comes later and is derived from it.
This is controversial, but even if accepted, that still leaves 4-6 centuries where the term referred exclusively to Travellers before the verb came about directly from that meaning.
Tinker as a verb is recorded in the sixteenth century.
And what does "directly from that meaning" mean when it's obviously about the action of mending and not anything else related to the group of people associated with the occupation?
Even assuming that this derived meaning was indeed the primary meaning of the word at the time, which is unclear (let alone the "exclusive" meaning):
Ti'nker. n.s. [from tink, because their way of proclaiming their trade is to beat a kettle, or because in their work they make a tinkling noise.] A mender of old brass.
---
> Who is Tom Snout?
ChatGPT said:
Tom Snout is a character in William Shakespeare’s play A Midsummer Night’s Dream.
He is one of the “rude mechanicals” (a group of Athenian tradesmen) who gather to put on a play—"Pyramus and Thisbe"—to perform for Duke Theseus’s wedding.
Occupation: Tom Snout is a tinker (a repairman of metal household goods, like pots and pans).
Role in the play-within-a-play: He is assigned the unusual part of “the Wall,” through which Pyramus and Thisbe whisper their love. He literally has to hold up his fingers to represent a gap in the wall.
Characterization: Like the other mechanicals, he is comically earnest and takes his role very seriously, though the performance ends up being hilariously clumsy.
Would you like me to also explain how his role contributes to the comedy of the play?
> Is he a Traveller?
ChatGPT said:
Not in the sense of being a “Traveller” as in a wanderer or member of a travelling people.
Tom Snout is just a local Athenian tradesman in A Midsummer Night’s Dream. Shakespeare introduces him as a tinker (someone who fixes household metal goods). He’s part of the group of craftsmen—Bottom, Quince, Flute, Snug, and Starveling—who rehearse and perform the little play for Duke Theseus.
So no, he’s not depicted as a traveller in the story—he’s rooted in Athens, more of a comic “ordinary worker” than a roving character.
Good thing it's been forgotten enough that almost nobody knows that, or cares it's the name of this company.
I think if anyone is offended by a word that is not used by anyone in that context, they're probably due for some self-reflection on what offends their sensibilities.
You are going on a very weird crusade in this thread. Literally have never heard this in my life used as any kind of slur and here you are arguing with everyone about it.
It is. But honestly, fuck that. It has a colloquial meaning that carries none of that baggage. Instead of retiring the word, why not make the racist definition archaic?
> colloquial meaning carries all of that baggage. You just weren't aware
The baggage—hell, all meaning in language—is carried by awareness. We don’t consider the word hostile racist because we’re not ancient Romans facing the hostis.
Maybe there is a cause to censor the word tinker in British English. What there isn’t is censoring it in American or international English.
I wouldn't touch this until they get serious about having real assurances that they're not going to access customer data without a real, justifiable reason. If Amazon gave themselves free reign to read S3 data it would be outrageous, this is basically the same thing.