RecipeNLG

Thanks for sharing not only the paper but the data used and also the models. We need more cases like this

I just got a fabulous recipe from the uncanny valley of cooking. Almost viable, with just a few mistakes that would have made it both impossible to follow and, if followed as…

Perfectly understood! We just cannot afford to run it on GPU in the backend so the generation takes up to 3 minutes which is terrible thing to do interactively :( You're invited…

It does take a long while though (I have been waiting for... I didn't pay attention but more than 15 minutes) so I can understand how using email would be the easiest solution…

https://recipenlg.cs.put.poznan.pl/generator linked in the Github page was looking good until > Second step is to fill Additional information. First of them - an email address. >…

Spicy Tomato Sauce Inputs tomato olive oil garlic salt sugar pepper Ingredients 1 (10 ounce) can crushed tomato 1 tablespoon olive oil 2 garlic cloves 1/2 teaspoon salt (to…

I tried ingredients for my current obsession: sauerkraut, just cabbage and salt. Naturally it did not reinvent sauerkraut. How many of those 2M recipes included a fermented…

Attempted to max out the ingredient list, and got a recipe that would probably feed 20 people[0]. Most impressive thing in my mind is it did correctly figure out it is a soup.…

I can't find any details on the license or copyright status of these recipes. It looks like a scraper was used (Scrapy). The only note I could find in the paper just says: >…

Ingredients lists are not copyrightable. The text and photos around them can be (as in a recipe book or blog). “A mere listing of ingredients is not protected under copyright…

FWIW, I'm no fan of giving my email either, so I tried to load the transformer on my own system. Would you mind to put a bit more, eg an example, than just following three lines…

"The dataset we publish contains 2231142 cooking recipes (>2 millions)" Where are the actual recipes? Sorry if I missed it, but I've been opening folders and couldn't find a huge…

I’m always surprised these datasets never have an example datum listed. It’s just click here to download all of it then see.

I got a recipe. My country is, as most countries are, metric. The recipe is in cups and pinches. It is software, so it is not difficult to add a question to the start of the…

I can't find any details on the license or copyright status of these recipes. It looks like a scraper was used (Scrapy). The only note I could find in the paper just says: >…

I’ve always wondered about this. Say you get the recipe and translate it using ML to another language but with fuzzy logic so it’s not identical to the original. And has added or…

"The dataset we publish contains 2231142 cooking recipes (>2 millions)" Where are the actual recipes? Sorry if I missed it, but I've been opening folders and couldn't find a huge…

IBM Chef Watson was quite good at it though, already five years ago. But I haven't found any similar service since it was turned off, what makes it so hard to replicate it?

yes, we got quite a long queue :D

This is great! I cooked recipes generated by GPT-2 for Thanksgiving dinner this year and it was surprising how a recipe can look good on paper but then come out in a completely…