Private LLM - Local AI Chatbot

First, I have an iPhone 14 Pro, and the app has layout issues. There's no option to save temperature for next launch. Opening llama 3 will crash the app. Shortcuts functionality is poor, as it doesn't allow custom system prompts only queries. Also, executing a shortcut function opens the app instead of quietly running in the background. Fixing these issues would instantly make this review a 5-star one

The app is shipped with RedPajamas 3B model, and it offers to download an unnamed 7B model, but if you refuse or close the app until the download is completed, it won’t offer it ever again until you completely reinstall the app. Model selection and download seems to be supported only on MacOS version, regardless of what written in patch notes and on the website. Sad, because i was really hoping to try Phi-3 on iOS.

Добавьте выбор любой модели - и будет вообще супер. Огромное спасибо за вашу работу. Отлично работает на iphone и mac. Единственное, что представленные модели очень плохо работают с русским языком. Еще раз спасибо за то, что вы делаете)

Your application is paid, but it should allow users to upload their own personal model via file. I use Llama3, 4Q, which works everywhere, while your model does not function correctly as it responds without understanding. You have only provided Llama3, 3Q. We request that you add a feature to select and upload a GGUF file that we have already downloaded.

Appena riuscirò a far essere le risposte meno casuali sarò ancora più felice 💓

Con 7b fa tanti errori, quando introducete 13b?

Here was my prompt: Request a LinkedIn recommendation from a friend who offered to do one for you. No more than 4 sentences. Tone should be very friendly and familiar and ask them to focus on these 2 things (make them placeholders), if possible. It wrote a recommendation that I would be giving for a friend, basically is not able to deduce or understand the meaning of the prompt. This happened repeatedly with prompts input. Put same prompt into ChatGPT - near perfect response. Tried a few different models including dolphin, LLama - same issues. Not sure if it’s in how to app approaches the LLM or some other factor, but it’s not as good as using ChatGPT or even Ollama on Mac.

My phone isn’t outdated as developer responded…. Default LLM is the only one that half way works but it isn’t worth it. It always states this at the end of everything. “The second paragraph is in English. 번역결과 넷, 제가 언어를 구한 2009년 3월 16일 LLM-데이터 수량을 5밀라 7밀라 8밀라 9밀라 10밀라의 요청은 "앞으로 어떻게 원하는 대지? 4단위 4단위 6단위 8단위 9단위 10 단위 11단위 12단위 13단위 14단위 15단위 17단위 18단위 19단위 20단위 22단위 24단위 27단위 28단위 29단위 30단위 등 참고해 주세요."

This app is pretty fantastic! It runs almost smoothly on my phone and I can use it anywhere without needing the cloud. Some additional features that would be cool are being able to browse the internet, interact with documents, get connected with a keyboard for rephrasing or correcting grammar on my phone instead of just on a computer. Also, there could be some improvement in user experience - smoother animations would be awesome. Keep up the good work!

Awesome app allowing you to choose which LLM’s you use for specific tasks. I think it’s pretty flawless the way it is great job on developing. I think it’s the best thing available today. Ultimate Wish list: functionality to simultaneously query multiple chosen large language models and aggregate their responses into a single answer by master LLM chosen.