Let's be kind each other, welcoming, you can solve your issues privately
Jean Louis
AI & ML interests
Recent Activity
Organizations
JLouisBiz's activity
Tool to GGUF conversion
Can't run in llama.cpp, wrong tensor shape


๐ฉ Report: Ethical issue(s)
Thanks for sharing this space!

hpcai-tech/open-sora-20-67cfb7efa80a73999ccfc2d5
โจ 11B with Apache2.0
โจ Low training cost - $200k
โจ open weights, code and training workflow
For both matters innovation and good example coming from great country of China.
How can you use multimodal capabilities?

Let's say this way, there is article, but no good reasoning in that article. It is talking about intuition without even defining it. And it talks zero of the fundamental principle of existence which is "survive". Maybe you could research what is the goal of mind: https://www.dianetics.org/videos/audio-book-excerpts/the-goal-of-man.html
Instinctive knowing is native to living beings only.
The new anthropomorphism of intuition given to computer doesn't make it so, just by writing an article.
Human mind wants to survive, and not only for oneself, but to survive as family, as group, as mankind, as all living beings, as planet. Some people are aware that planet must survive, like Musk, so he builds rockets for Mars, while other people can't understand why. Though the better survival level we seek, the better we do over long term.
Computer doesn't want to survive, it is tool like a hammer. It has no intuition, it has no survival, thus has no instincts.
You can of course try to build data and try to ask computer to act upon such data, which in general, majority of models already do. They are giving probabilistic computations, but know nothing about it. Intuition is human and description of it has been already built in into the LLMs. If you wish to improve it, you are welcome.
However, I don't see anything revolutionary here.
LLM is reflection or mimicry of human knowledge.
If you give it some operational capacities such as to move around, to target people in the war, to control the house and business, it is going to do by the data it has been given, and it will do disaster randomly, just as it gives random nonsensical results from time to time.
Gemma License (danger) is not Free Software and is not Open Source:
https://gnu.support/gnu-emacs/emacs-lisp/Gemma-License-danger-is-not-Free-Software-and-is-not-Open-Source.html
Let's not forget the meaning of Free Software models like DeepSeek and Qwen, Mistral, and that Gemma is proprietary, Freeware, but not Open Source
Is it possible to train it on a single 3090 using LoRA on PEFT?
I have a Dmenu script to switch between the models. https://gitea.com/gnusupport/LLM-Helpers/src/branch/main/bin/rcd-llm-dmenu-launher.sh
I just click and then choose the model from the menu.
You mentioned switching modes and switching model. Now, what do you mean with switching the modes?
And finally, you can just talk to your model and ask it to give you the shell script or any other kind of programming code to help you switch the mode.
The important is that you have defined how to run the model by some command, and then you can put all those commands together in a list, and then you find a way how to switch modes.
I like speaking, even now I'm speaking and getting this comment in text. So that means I could basically speak and have my computer intercept the speech before it comes to any model. And then I can use embeddings to basically recognize if I have given some command. You can even use the simple script recognition. Like you could use string recognition. And then based on your spoken command or maybe the text which you are entering, then you could switch the mode or switch the model.
I have tried with a small data set and it never finished the training. It took really, really long time. I don't know how long. Maybe one hour. I don't know. I was just watching how it is in the stage one, but it was very, very long time.

Tutorial Link : https://youtu.be/ueMrzmbdWBg
Squish Effect LoRA arrived to Wan 2.1. Wan 2.1 is the truly State of the Art (SOTA) Open Source video generation model that supports Text to Video (T2V), Video to Video (V2V) and Image to Video (I2V). Now our ultra advanced 1-Click Gradio application supports LoRAs and today I will show you all the new developments to our Wan 2.1 all in one video generation Gradio App. We have added so many new features since the original Wan 2.1 step by step tutorial and we continue to improve our App on a daily bases with amazing updates.
If you want to have Squish it: AI Squish Video Art locally for free forever, our app and Squish LoRA and Wan 2.1 is all you need. Watch this tutorial to learn all. Moreover this tutorial will show you majority of the newest features we have implemented with non-stop working for 10 days.
Hopefully many more updates coming soon.
I have just tried it and it could not read the map. It is very inaccurate with words which are clearly digitally written on the map. There are many different use cases for your use case. It is good for mine, it's not.
The big difference between Google Gemma and Qwen models is that Google is not open source, it's not free software, while Qwen is truly free software, free as in freedom.
Comparing those models from quite different categories is not right.
Gwen is not limiting commercial users, not even largest companies or governments, while Google does.
Comparison makes no sense.
