Stream Notes
Catch Up
Computer didn’t work this morning
Spent all day troubleshooting/fixing
Hopefully it works again…but it’s back in my room and it’s hot
Doing
Let’s figure out how to save these embeddings
I have photos, Vespa works with embeddings, can I use the photos
We got raided HanamiYoshino
Egg tart, Chinese
Cream Cheese Pound Cake
Ingredient faithfulness good
Cooking prep not close
Cuisine difference clear
Interesting yes
Good suggestion
Classic Gougeres
Ingredient faithfulness ok
Cooking prep not close
Cuisine difference clear
Interesting yes
Ok suggestion
Warm Chocolate Tart
Ingredient faithfulness very good
Cooking prep very close
Cuisine difference not clear
Interesting no
Ok suggestion
Profiteroles with Caramel Sauce
Ingredient faithfulness very good
Cooking prep not close
Cuisine difference clear
Interesting yes
Good suggestion
Toasted Pecan Pie
Ingredient faithfulness very good
Cooking prep very close
Cuisine difference clear
Interesting no
Ok suggestion
From Chat / Derail
#hotchickencurry do foodguessr and take results through this program
#hittaus
First point: for personal projects try using JJ, jujitsu version control , over git. To learn read Steve Klabnik book about JJ . Thank me later.
Super cool, sounds really helpful!
Here: Multimodal point: have you tried anything fun lately? Here’s fun multimodal thing to try. Install ollama if you don’t have it already. Install mini-cpm model. (ollama pull minicpm but prolly gotta check the actual model name). Then try command line: ollama run minicpm “what’s the food and probable recipe that’s producing this dish in image path_to_picture.jpg ”. Also, I have no idea if it works with food.
Socials
Related
