Видео

I learnt Rust and became proficient at it over the course of two years and I totally agree with you. It's software development/engineering that is inherently complex, not Rust. Rust however, unlike other languages, doesn't hide that complexity from us. I see the compiler like a strict but helpful computer science teacher that doesn't accept blatantly bad code. I learnt so much more about software by learning Rust, especially as I asked ChatGPT to explain me the tricky concepts I didn't instantly grasp in Rust.

i'm a hobbyist programmer and i love peeking into different languages. i admire the determination of Rust devs to combine C with Haskell, but i think that the result is underwhelming both in comparison to C and to ML style languages. like C is so good for low-level precisely bc it doesn't encourage RAII (which can be very slow for a large number of small structs), and because it doesn't have phantom types and traits and all that bs -- it's raw pointers and offsets. functional languages on the other hand have a lot of stuff to manage traits and other type frankensteins (which only offer some basic safety checks, btw). C's greatest weakness is lack of proper compile-time programming. though Rust has a powerful macro system, it's very unpleasant, and i can't see why it couldn't be more like in Zig or Julia.

brilliant

I so so so much agree.

Rust is complicated

Naa Rust is complex , I can justify, developer want to leave C bcz there you have to manage memory by your won , and same goes to rust too , it's you have to think about owner ship model , so in simple words you are doing same thing but in diffrent language and different way , Rust is safe but it's comes with a cost of high complexity non-readable

"It's software engineering that is complicated. Rust doesn't allow you to avoid its complexities." ...I think I just added something to my vocabulary.

I'm impressed how fast you scrolled through the article unlike some other youtubers who'd shout, scream and regurgigate every other sentence just to add nothing.

Rust definitely is not for people who are not willing to learn difficult stuff. I don't see a point in lowering expectations here. Is Rust worth it anyway? It was for me, but that doesn't mean it will be for everyone. What I have seen in my career is, that many employers and superisors think, that when it looks like it works, then they could have gotten away with less work and want to ensure not to repeat that mistake in the future, then Rust is rather like a land mine. If your goal is to create high quality software and your superiors have your back, Rust is the main contender.

I cant agree more, this is such a modern and amazing take. Exactly why I code websites in axum

I was wondering what is happening in the background and cannot find correct information. This video exactly what i was searching for.Thanks !

Indeed. They are just recreating the k8s. I’ve tried interacting with them on X about various ego issues they have. Shockingly I never get a response lol

Is k3s good for production environment?

thank you for the video , how do we limit the knowledge base only limited more to the pdf uploaded. .

When other videos?

Can I use local models in bionic-gpt?

Hi, great work on Bionic GPT. Tried installing it locally in. my macbook but I am encountering errors when using local Llama model via Oomla. It gives me an error 111 refused connection error. Can you please point me to the resources which may help me sort this out? Thank you.

Seems like it cut before the end. Any plan on adding the rest ? Thanks !

does it work with ollama?

Thanks for this, the system looks amazing !!! The questions I have are - how to set this up on a mac or windows? docker? - can it run on a single machine? I'm not overly familiar with kubernetes but can a cluster just be one machine ? Many thanks in advance !

Is it possible to use bionic GPT with Azure OpenAI Service?

Great!

So how do we add models? 🤔

Hello, I am new to this but I was wondering can this chat run on 16GB Ram/ i5-10300H CPU @ 4.5 Hz/ GTX 1650 4GB ? thank you in advance

Well done, you helped me a lot! I have a question about 0:58 - How to get more models?

Is it possible to do Multi-document summarization with BionicGPT? I would imagine an agent on top of each document and one agent on top comparing the outputs of all the agents and feeding it to the LLM model. Or is BionicGPT primarily set up to find the most relevant source of information based on the prompt you give and then gives a specific answer together with the LLM?

Haven't been able to get the embeddings-api service to run with GPU support. Any advice? I've installed the Nvidia toolkit for docker, and GPU works for the llm-api service.

for windows ??

Let's say that i've got a bladecenter that has 16x 16-core 2.10GHz processors just laying around... How would/could I cluster the hardware together to run something like this?

Thank you very much, Ian, for the amazing work you guys are doing. I can’t wait to try it out.

is there an installation video, step by step guide on hot to download on each operating system,. such as windows, or mac.

Hi @Ian I have just installed as a docker container and opened the UI at localhost. When I submit a prompt it shows the following error on the console. I created new prompts and tried but no luck. Please help object has no attribute 'request' oauth2-proxy-1 | 2024/01/05 18:27:42 httputil: ReverseProxy read error during body copy: unexpected EOF oauth2-proxy-1 | 2024/01/05 18:27:42 httputil: ReverseProxy read error during body copy: unexpected EOF oauth2-proxy-1 | 2024/01/05 18:27:42 httputil: ReverseProxy read error during body copy: unexpected EOF

Thank you been playing around with it.

I am incredibly excited about your work. Great job, so far!

Does this work on a Macbook M2 Pro? I tried every way to install it but failed. Getting this error Error response from daemon: could not select device driver "nvidia" with capabilities: [[gpu]]. The only way that works to install and run is via the "Installation (CPU)" route but then no responses are generated by the LLM, I get an error with every prompt. I'd appreciate any help. I'm not very technical but can do some basics. It would be so great to use this.

Hi Ian, I tried it out and it runs fine with 16gb RAM. However I am having trouble setting up a new LLM. I tried adding Mistral 7B but it doesn't work. I went through the documentation on your website but still facing the same issue. Can you please assist for the same??

how many GB of RAM is needed? I tried to install it on a 8GB RAM notebook, could get into the UI but then the models did not respond Thanks in advance and thank you for making it possible

Spectacular .. Genius Mate!

Looking forward to capability extending to MacOS and M hardware.

if my notebook is 8GB of RAM.. doesn't this work? Because when i start a chat it says "Processing prompt" and the loading icon occurs.

Great work guys! Id love to connect with with autogen. Is this possible?

This project will go big. Great work.

can you pls show/explain how to integrate with other top open source models like OpenChat3.5 i get Error occurred while generating

getting error Error occurred while generating. in docker logs getting 2023-12-02T14:29:51.838799Z ERROR axum_server::errors: response="status = 422 Unprocessable Entity, message = query returned an unexpected number of rows" help please

It works great! Would be nice if i can switch to gpu instead of cpu

Great! esp for small teams to deploy. thanks

wow. I shall learn to do that too! Lots of fun.

This looks awesome. Thanks for making it open source, I am going to try it out with my organization.

love it