February 17, 2024

Team Status Report for 2/17/24

Risks:

Meta Llama 2 LLM has some trouble giving a variety of recipes and at first tried to use as many ingredients as possible. We will test more specific prompts to see if it will work as intended. Another approach can be to try to fine-tune with a specific dataset of recipes.

Changes:

Debating whether or not scanner should include a screen on device or if visual debugging feedback will be presented in the app. May be performance driven.

Progress:

Completed App UI design
Scanner CAD mock up
Design Review Presentation
Obtained the RPi Zero 2W and RPi camera, but found that the cameras need a specific adapter for the Rpi Zero 2w, so will order
Successfully downloaded Meta Llama 2 LLM and started testing prompts

Additional Questions:

A was written by Kelly, B was written by Michael and C was written by _____.

A. Sous-Chef will meet the need of public health and safety because it will encourage people to make use of their available ingredients from the recipes that the app will suggest. The convenience of having a recipe with ingredients you already own will encourage healthier cooking habits, especially for those who might be overwhelmed with cooking. This can promote a balanced diet, as opposed to ordering unhealthy food for delivery or at a restaurant. Additionally, by sending push notifications of expiration dates and making sure recipes are recommended so that users can finish the ingredients before they spoil, Sous-Chef helps prevent the consumption of expired food, reducing the risk of foodborne illnesses.

Some people might be anxious about needing to use their produce before it goes bad or they might forget about ingredients they already have and later feel guilty that the food will have to go to waste. Sous-Chef supports psychological well-being by alleviating concerns about the freshness of your ingredients. Planning what to cook can also be a mental burden since people need to make time to check what ingredients they already have and do not have for a specific recipe. Searching for and deciding on a recipe will also take time, especially for those who are indecisive. This app will remove that mental load from the users and make sure that the food they are eating is safe and a recipe can be generated quickly without much thought or effort.

B. Sous-Chef will consider social factors in two main ways. First, we’ll be cognizant of dietary restrictions (such as being vegan, gluten free, etc.) by allowing users to select any combination of predefined options before prompt engineering to request a dish. By having the option to create dietary sensitive meals, we’re being inclusive to groups that might typically be wary of meal generation/suggestion apps.

Additionally, because Large Language Models are trained by scraping the entire internet, there will be a wide variety of cuisines that Sous-Chef will have been trained on. This means that if people tend to have a lot of ingredients related to one culture, Sous-Chef will be able to recommend meals that don’t stray too far from home. However, if people want to experiment, there’s always the option for Sous-Chef to recommend a dish from another culture or even create fusion dishes. This will make people more aware of other cultures, and may even drive some business to local ethnic diners if people enjoy a dish recommended by sous-chef.

Reduce the need for frequent grocery shopping because you have a list if your fridge’s inventory (being sustainable)
Minimize wasting food (wasting money) because users will use it before it goes bad
Minimize going out to restaurants/ ordering food since you can cook with what you have and not spend time trying to find recipes/ meal planning