PaLM2 Translations with Custom Glossary

When released, PaLM 2 not only surpassed other LLMs in translation abilities between multiple languages but also demonstrated significant improvements over Google Translate (see the Technical Report by Google). This example allows you to easily change out the languages and translation text via variables to explore this and compare with other LLMs. You can also add your own TSV or CSV glossary in the format spec'ed by Google Translate. This example run (or runs like it) can be imported into the Translation Recipe to compare with traditional translation tools!

2y ago

👩‍💻 Prompt

🧠 Language Models

💪 Capabilities

🛠️ Developer Tools and Functions

Variables

text_to_translate

string

Hi, I'm Sean, one of the founders here at goowy. And I wanted to give you a demo today of how to use one of our recipes, a face in any scene. So just go ahead and click on this. And so the scenario that I'd like you to imagine with us is let's imagine, you know, rather than trying to do stuff with Elon Musk, we are working for a travel agency and we've got a spokesperson and we want to show that person basically all over the country in a variety of locations. Which is expensive because you got to send that person out, you got to do a photo shoot. And so what I'll do is I can just drag in a picture. So I have here Lana, AI generated spokesperson that I've drafted right here. And then I can just take any other quote. So I'll just copy in another one right here, which is basically a beautiful woman to the grand Tetons in Hawaii or Wyoming and award-winning landscape photographer. And then you literally just do those two and then hit submit. What happens now is we're looking for this face and then we basically go ahead and we craft out or mask out that particular face. We readjust it and then boom, we've got this picture of Lana hanging out at the grand Tetons. You can go ahead and download that and we can change anything else. So if we want to say, well, no, she's not at the Grand Tetons, you know, she's in something like Maroon Bells, another famous American destination that I've never been to. We can put that up and then you'll notice the same thing happens. In about nine seconds, we do the same face repositioning, same face masking, we put that in and then we do a little bit of cleanup at the end with JPF CAD. And so this is an example of a four stage recipe that hopefully anybody can go ahead and use just by uploading a picture of a person and any text they want. So we hope you enjoy it and we look forward to seeing what you make with it. Thanks so much.

glossary

string

target_language

string

source_language

string

⚙️ Settings

Run cost = 1 credits

With each run, you agree to Gooey.AI's terms & privacy policy.

PaLM 2 Chat • Google [Redirects to Gemini 2 Flash]

Hej, jeg er Sean, en af grundlæggerne her hos Gooey. Og jeg ville gerne give dig en demo i dag af, hvordan du bruger en af vores opskrifter, et ansigt i enhver scene. Så bare gå videre og klik på dette. Så det scenarie, som jeg gerne vil have, at du forestiller dig sammen med os, er, lad os forestille os, du ved, snarere end at forsøge at gøre ting med Elon Musk, vi arbejder for et rejsebureau, og vi har en talsmand, og vi ønsker at vise den person i det væsentlige over hele landet i en række forskellige lokationer. Hvilket er dyrt, fordi du er nødt til at sende den person ud, du er nødt til at lave en fotooptagelse. Så det, jeg vil gøre, er, at jeg bare kan trække et billede ind. Så jeg har her Lana, AI-genereret talsmand, som jeg har udarbejdet lige her. Og så kan jeg bare tage et hvilket som helst andet citat. Så jeg vil bare kopiere et andet ind lige her, som i bund og grund er en smuk kvinde til de store Tetons i Hawaii eller Wyoming og prisvindende landskabsfotograf. Og så gør du bogstaveligt talt bare de to og trykker derefter på send. Hvad der sker nu er, at vi leder efter dette ansigt, og så går vi i bund og grund videre, og vi laver eller maskerer det pågældende ansigt. Vi justerer det, og så, vupti, har vi dette billede af Lana, der hænger ud ved de store Tetons. Du kan gå videre og downloade det, og vi kan ændre alt andet. Så hvis vi vil sige, nå, nej, hun er ikke ved de store Tetons, ved du, hun er i noget som Maroon Bells, en anden berømt amerikansk destination, som jeg aldrig har været på. Vi kan sætte det op, og så vil du bemærke, at det samme sker. På omkring ni sekunder laver vi den samme ansigtsompositionering, samme ansigtsmaskering, vi sætter det ind, og så laver vi en lille smule oprydning til sidst med JPF CAD. Og så er dette et eksempel på en firetrinsopskrift, som forhåbentlig alle kan gå videre og bruge blot ved at uploade et billede af en person og enhver tekst, de ønsker. Så vi håber, du nyder det, og vi glæder os til at se, hvad du laver med det. Mange tak.

🐞 Debug

How to Use This Recipe

🙋🏽‍♀️ Need more help? Join our Discord

PaLM2 Translations with Custom Glossary

👩‍💻 Prompt

🧠 Language Models

💪 Capabilities

🛠️ Developer Tools and Functions

Variables

Response Format

Max Output Tokens

Creativity (aka Sampling Temperature)

Answer Outputs

How to Use This Recipe

Related Workflows

Create a perfect SEO-optimized Title & Paragraph

Profile Lookup + GPT4 for AI-Personalized Emails

Agent Builder

Lipsync with Text-to-Speech