Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

news.ycombinator.com

Post reply on HN
Print view View original

21–30 of 50 posts

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#21

Post by Melting_Harps »

danott wrote: post #7
isusmelj wrote: post #4

Hi George, Oscar and Dan. Congrats on the launch! Nyckel sounds super interesting. It looks like you mostly focus on classification tasks. Any plans to also support other tasks such as object detection or image segmentation?

Thanks. They're both on the roadmap; in fact we've got a handful of users in a private beta for object detection; let me know if you're interested in getting in on that!

I'd like to hear more how I could contribute to the beta, I'm studying AI and ML after a career in fintech and would like to see how things are done to train these models first hand.

I'm playing with Nykel right now and it looks pretty straightforward with a clear and simple UI.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#22

Post by isaacimagine »

danott wrote: post #13
blululu wrote: post #9

This is very cool. A few quick questions: 1.) Would it be possible to buy the model and integrate/host it on my own machines? 2.) Would you consider making solutions for embedded ML in the future?

We do think "model export" is important, but we're still getting our heads around how to do it in the most non-ML-expert friendly way. We don't think the persona we're building for wants a weights file dropped in their lap. What output / format would be ideal from your perspective?

Maybe a cross-language library that takes a binary weights file (with embedded model information) and exposes an interface similar to that of the web API? Or a local lightweight version of Nyckel that one can run on their own infrastructure (that exposes the same REST API)?

Just spitballing here; these two would be the most convinent for the use-cases I have in mind.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#23

Post by blululu »

danott wrote: post #13
blululu wrote: post #9

This is very cool. A few quick questions: 1.) Would it be possible to buy the model and integrate/host it on my own machines? 2.) Would you consider making solutions for embedded ML in the future?

We do think "model export" is important, but we're still getting our heads around how to do it in the most non-ML-expert friendly way. We don't think the persona we're building for wants a weights file dropped in their lap. What output / format would be ideal from your perspective?

I was thinking of something like an ONNX file or something that can easily slot into different runtimes.

Makes sense that this would be less beginner friendly so maybe you're correct that this is a P2 feature.

I guess I was thinking more in terms of pricing models and scaling up a service which is obviously a complicated decision for a startup so I'm not really sure what makes sense here. My rationale for wanting to buy/rent the model is that as a service scales it becomes increasingly important to own the model and the hosting. One of my concerns with building on top of a service like this is that it will potentially reach a chokepoint in the future. In general training a model is expensive and unique but hosting it is a commodity service. This will incentivize customers to use the service when they are small and then drop it when they grow to a certain size which is not necessarily ideal for either party.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#24

Post by beijbom »

Melting_Harps wrote: post #21
danott wrote: post #7

Earlier quoted context omitted.

Thanks. They're both on the roadmap; in fact we've got a handful of users in a private beta for object detection; let me know if you're interested in getting in on that!

I'd like to hear more how I could contribute to the beta, I'm studying AI and ML after a career in fintech and would like to see how things are done to train these models first hand. I'm playing with Nykel right now and it looks pretty straightforward with a clear and simple UI.

Thanks! I'd love to chat more about your use-case and how we can help you. Drop me a line at oscar at nyckel dot com to set something up.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#25

Post by saintarian »

isaacimagine wrote: post #22
danott wrote: post #13

Earlier quoted context omitted.

We do think "model export" is important, but we're still getting our heads around how to do it in the most non-ML-expert friendly way. We don't think the persona we're building for wants a weights file dropped in their lap. What output / format would be ideal from your perspective?

Maybe a cross-language library that takes a binary weights file (with embedded model information) and exposes an interface similar to that of the web API? Or a local lightweight version of Nyckel that one can run on their own infrastructure (that exposes the same REST API)? Just spitballing here; these two would be the most convinent for the use-cases I have in mind.

Agreed that they would be convenient. We are looking at both those options. There are devils in the details like seamlessly taking advantage of available hardware acceleration.

Would love to talk more about your use case so we prioritize the right things for model export. Drop me a line (george at nyckel dot com).

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#26

Post by danott »

blululu wrote: post #23
danott wrote: post #13

Earlier quoted context omitted.

We do think "model export" is important, but we're still getting our heads around how to do it in the most non-ML-expert friendly way. We don't think the persona we're building for wants a weights file dropped in their lap. What output / format would be ideal from your perspective?

I was thinking of something like an ONNX file or something that can easily slot into different runtimes. Makes sense that this would be less beginner friendly so maybe you're correct that this is a P2 feature. I guess I was thinking more in terms of pricing models and scaling up a service which is obviously a complicated decision for a startup so I'm not really sure what makes sense here. My rationale for wanting to…

Yep. The pricing model does basically break down for model export, but I think there's a solution there. Or, said another way, if we could make it really easy to do then there's an adjacent business we could move into.

In terms of keeping customers as they grow, our view (hope?) is that these models will be continually updated because of new annotations on their end, and from new training techniques on ours. And that concept of continuous improvement will push people toward a SaaS model.

When you say chokepoint, are you referring to cost, or latency, or something else?

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#27

Post by colincooke »

Interesting product strategy, but I can't help laugh at some of the product examples like "barcodeless scanner" or "quality inspection" (I can only comment on imaging as that's my background). First of all the idea of replacing a perfectly functional barcode scanner with an ML model is not a great sell (having how much I pay at the register dependent on the lighting at the grocery store is not going to be a fun time). Second, ML models are great when failure is OK at a specific rate, none of the image classification examples shown have that characteristic.

While I'm sure the founders are competent and understand these limitations, it's unfortunate that they've chosen such flawed examples to show off on their home page.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#28

Post by danott »

colincooke wrote: post #27

Interesting product strategy, but I can't help laugh at some of the product examples like "barcodeless scanner" or "quality inspection" (I can only comment on imaging as that's my background). First of all the idea of replacing a perfectly functional barcode scanner with an ML model is not a great sell (having how much I pay at the register dependent on the lighting at the grocery store is not going to be a fun time)…

Actually, all these examples are straight from our userbase. I can't speak to how successful they're being in their own businesses, but they seem happy with the ML. The 'barcodeless scanner' in particular is about scanning bulk foods that don't have barcodes, and it seems to work for them.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#29

Post by gamerDude »

This is a very interesting product, and since I have some data that could really benefitted from this, I tried it out.

I went through the upload process. But then I don't really know what to do from there. I tried some filters. I went to the invoke page, but I had no idea what invoke does or what the example output is. (Eventually figured out that I can just put text in the invoke and run it). All in all, there are a bunch of things that I don't really know what they are. I was a statistician before ml became popular, so I understand the underlying premises, but none of the modern language.

I would also really have liked to been able to filter by say, if the confidence level is over 80%, how accurate is the model. Because then I can say, well, if we use this, I can knock out tons of work at the 80% confidence rate and then just manually work with the rest.

I'm also not sure if you are seperating training/test data. All in all, looks nice, it was very easy to get started, but I'm a bit lost on what to do next and I'm having trouble judging how useful this will be to me and if I should invest more time.

Re: Launch HN: Nyckel (YC W22) – Train and deploy ML classifiers in minutes

#30

Post by saintarian »

gamerDude wrote: post #29

This is a very interesting product, and since I have some data that could really benefitted from this, I tried it out. I went through the upload process. But then I don't really know what to do from there. I tried some filters. I went to the invoke page, but I had no idea what invoke does or what the example output is. (Eventually figured out that I can just put text in the invoke and run it). All in all, there are a…

Thanks for trying us out and for the feedback! I agree that our filters are a little confusing right now and we're working on fixing it. In the meantime, here are a couple of filters you could try:

- To see all cases where the model disagrees with your annotation: Function Output = Disagrees, Desired Output = Any.

- To see the least confident predictions from the model: Function Output = Any, Desired Output = Any, Sort By = Least Confident Prediction.

Your idea us helping you pick a confidence threshold is a good one. We'll get that into our near-term roadmap.

We use a technique called cross-validation to seperate training and test data. We have that documented here: https://www.nyckel.com/docs#cross-validation

Post reply on HN