Smart Second Brain for Obsidian(Free & Offline)
HTML-код
- Опубликовано: 5 июл 2024
- In this video, we'll talk about how you can integrate local LLMs within Obsidian to chat with your notes and chat with the AI without any third party services like Claude AI, & ChatGPT. We'll be using the Obsidian Smart Second Brain plugin to help us achieve the setup.
OLLAMA: ollama.com/
Smart Second Brain: github.com/your-papa/obsidian...
//🔗 LINKS
💌 Newsletter: obsidianninja.com
⬇️ Obsidian Life OS: beingpax.gumroad.com/l/Obsidi...
//🙇🏼 ABOUT ME
Writings & Articles: / beingpax
Twitter: / joshipax
Digital Garden: garden.prakashjoshipax.com/
//🫶🏼 SUPPORT ME
On Patreon to Get Exclusive Benefits: / beingpax
BuymeaCoffee: www.buymeacoffee.com/beingpax
//⏰ TIMESTAMPS
00:00 Introduction
01:26 Setup & Configuration
07:46 Exploring Main Features
10:45 Best LLMs
13:17 Best Combination
Thank you so much!! I was about to start a brand new vault with two years of research because I was concerned that I had too many notes. This has saved my life! Thank you so much. I really appreciate this.
Thank you Prakash for your videos. You have been providing immense value to the community.
Thank you for this awesome and insightful video, man!
Thank you so much for creating this video. Every aspect has been helpful.
Fantastic Content, my friend. Keep it up!
By far the best ai use I have seen so far in obsidian
This is wonderful!
Love athe tutorial. When I have my new pc i will surely go into it.
Thank you for the comprehensive plugin presentation!
I'm curious about the parsing notes into vectors process: does it take into account any linking or tagging, or does it parse only text information?
What do you think?
thanks for showing - nice video 👏
Ollama is pronounced "Oh Lama".
Thanks for the great video!
How is this model with languages other than English?
Pax, your video quality is getting better and better. What are you using for your video-Editing?
Thanks ❤️. Right now, I'm transitioning into Final cut pro, because of how fast and reliable it is.
I'd love to know when they're going to add the Llama3 model. Also, could you make a video explaining the differences between Nomic and MXBai? How do they work, and what is best for each case? Is there a way to use LMStudio instead of Ollama? Liked and subscribed. Amazing video
It´s already possible
I'm using llama3 for both embedding and generation. And it works much better that way than the default embedding models.
Wow. That looks useful. Did you try that with other languages than English?
just english and german, when llama3 can tak to spanish, there is a way to change assistant language?
next level...
Somehow, when I attempt to obtain sources from which the bot gathers its information, it only provides me with numbers, e.g., (Source: 2). Then, when I click on 2, a new note with 2 as the headline appears. I have the feeling it does not take information out of my volt.. Where am I going wrong?
Thank you in advance!
Hello Prakash,
Firstly, thank you for your videos, which are always very professional and
comprehensive. I need your help. Following your guide on Linux, I
installed the Obsidian plugin, configured it, and downloaded Ollama and
Gemma. However, in the Smart Second Brain section of Obsidian, I am unable
to configure the AI to function correctly. The aspect of my setup
that is much less complete and detailed (only for point) than the one you show on macOS. In
particular, I am missing the "start Ollama" command.
When you open the Smart second brain pane, you have the option to set Ollama origions to enable streaming responses. Follow those instructions and it will probably solve.
Whats the exact problem you are having?
how could i connect it if my ollama is inside the wsl of my windows?
thanks
Thank you for making the video on AI......I strongly believe that A.I will dominate the PKM area.
If you have .pdf files in our vault does this plugin index the PDF files? If so, does that significantly increase the indexing time? Also does it use the information in the PDF to answer questions?
It doesn't support pdfs, yet.
Is there a plugin that can use A.I. to create and edit notes as well? This would be a big leap forward as far as efficiency of note taking goes!
My idea was to take an idea explored with an LLM then have the LLM summarize the the conversation and conclusions, or you coupd write it in your own words, then the LLM can automatically generate the nodes. The original conversation could be saved in a separate text file that doesn't bloat your vault, which can be linked to in the note. It shouldn't be too hard to automate all this, but I'm not a programmer, or else I would love to make it a reality.
sadly it does not show up after installing. must clash with some of the other plugins.
Can i run this in Polish language by any chance? currently only english is avalibe i wonder if there is any LLM that works with Polish
What's your hardware spec? It seems to be taking forever to index as well as generate answers on my ryzen 5 laptop
I'm on Macbook Pro M2. Some local models take way long time. I find the mixtral model with mxbai works the best.
What specs (ram / processor) do U need on your computer for running a local LLM like the new llama 3 7b ? I have a mini Mac with 8gb
8 GB will HAMMER your swap (the internal SSD, which I gather is only 256 GB) and end up destroying the SSD.
@@HiltonT69an interesting detail to this whole story:)
My vault is over 3million words across 2k+ files, all geared towards one project. I was excited to use the Co-Pilot plugin because it advertises "vault mode", but was quickly disappointed when its default reference amount was only 3 notes/files at a time (lol), and can only stretch to 10 but gives a warning that it will prolly screw up the responses.
I want to communicate with my vault as a whole for perfect macro context, but it seems my case use is still not possible with current Ai? SmartConnections doesn't seem to be any better, or am I mistaken?
This local solution might work for you. Try an embedding model like bge + llama3 for generation. It'll take a LONG while the first time though
Hello sir, I have encounter this "Error: listen tcp 127.0.0.1:11434: bind: Only one usage of each socket address (protocol/network address/port) is normally permitted.". I have taskkilled the PID but it's not worked, another PID has come. Could you help me ?. Do I have to install the python or anything else ?
have you found a way to exceed this problem?
What LLM should I use for a different language (German)?
Schwierig
do you know where on the computer these models are stored? I have trouble finding the files...and they are huge so after trying some models you might want delete them
Did you happen to learn how to delete the models?
I love that it is free and private. What do U think about other AI plugins like smart connections?
The main differentiator is this: It supports open source models. While Smart connections uses OpenAI APIs. And you have to pay for it. Smart connections also has license model for which you have to pay to access some features.
@@beingpax thanks for explaining👍 just one more question: if I add a note after Indexing, do I have to do a new index, and does it recognise, just the new note or do I have to index the whole vault again and where will it be saved?
Not 100% sure, but I think it will only index the new updated notes only.
❤❤❤❤❤
I am getting error message "Failed to initialize Smart Second Brain (Error: TypeError: Failed to fetch)" ?
same problem here
All these tests make me think that Smart Connections does a much better job of this...
my ai why is slow response ?
Благодарю за полезную информацию. Thank you for your assistance.
Yt sucks hard and removed my comment about using Llama 3, which is possible to do!
How is it and whats the system requirements to run these stuff
From my point of view, I think the Copilot plugin is relatively better than this plugin.
the llm is pronounced
“OH-LA-MA”
c’mon ma’an !!!!!