Snappy skills by Micro Learning useful topics on tools and technology with AI. AI for Enterprise business applications is where we specialize! Check www.LevlUp.AI for services
We make videos about AI tools, comparisons, product reviews and opinions on AI tools as well as providing step-by-step guides, tutorials, and expert advice.
Whether you're looking to explore AI, common software, master a new gadget, software or enhance your experience using AI we've got you covered. Our goal is to make content accessible to everyone, no matter your skill level.
Tools interest in the collection: - VertexAI - ChatGPT - Enterprise AI - Google Cloud Platform AI - Data Analysis & Data Development
We are not affiliated, nor impersonate channels and the content created is the work of LevelUP_Plus. Where applicable works that reference the original artist, software, or content will be referenced.
I wish people would consistently call Visual Studio Code or VSCode instead of just Visual Studio. There's a different program called Visual Studio, the giant IDE with graphical form designers. I clicked on this video hoping to see one about Visual Studio, not Visual Studio Code.
Is it that Chinese data is not well supported? I've noticed that when a notebook is filled with a large amount of source material, NotebookLM seems to favor the first data entry added. Even though the nth entry added later contains the information I need, the search tells me it can't be found... Is it just a fantasy to try to find what you want from 50 sources? It seems that when there's only one data source in a notebook, the question-answering performance is best. Adding one more source also seems okay, but after seven or eight sources, no matter how I ask, it only cites sources from the first one or two entries.
hey after tuning my data it is successfully tuned but when I test my model and try generating prompt I am getting this error ERROR Cannot read properties of null (reading 'length')
hi i have created invoice parser processor and tried with api request but i got the raw it doesn't giving response with lable and value which i trained
I finetuned gemini 1.0 pro for text classification. While finetuning i also provided a validation dataset. It showed some metrics called "eval_fraction_of_correct_next_step_preds" to be around 0.93 at the end of training. I am assuming it as the accuracy. But when i run the same model for my validation set locally and caluculate accuracy it is coming around 0.77 Then what is the metric that is shown in the vertex ai studio while training?
Hi there, thanks a lot for this. Trying to test somethign with it, but I get the "Failed to create pipeline job. Error: Permission 'aiplatform.metadataStores.get' denied on resource" error. I think it has to do with the service accounts and permissions. For me it only shows the compute engine default service account. I've gone to create service accounts but over there the instructions I've seen ask me to download a key, and that's no longer allowed. Could you please nudge me in the right direction on how to get the service account setup properly so that I could select it here? Thanks!
Hi, if you haven't already you might need to go through the Google Cloud Setup steps to configure your environment, so click the Google Cloud Setup, and go through the wizard. The user access and service accounts are provisioned in there.
Hi, thanks for watching. It is linked to your GCP platform account, Gemini Code Assist is available to try at no cost until July 11, 2024. cloud.google.com/products/gemini/pricing
Hey Thanks for the tutorial. Seems we have a common problem in the comment section: 403 Error unauthorized access. Using the resources below I wasn't able to still access it outside google AI studio. I want to integrate the model in my application. Could you guide us or maybe make a video of it ? . Thanks
Hi thanks for the comment, I have the GCP environment setup - so check the Google Cloud Setup and follow the wizard to enable the service accounts and the access permissions. I will have a look at what I specifically did.
Hi Thanks for Watching, yes it can - see some of the videos in my channel, to get some ideas. But you ultimately do prompt engineering, and build a RAG type of chatbot.
You have fine tuned right, if you ask questions which are not in fine tuned data (excel data) will it answer ? Ex: explain about database? Now this finetuned model will answer correctly
I dedicated trying to understand their totally feature bloated platform for a full week and got sick to my stomach with how unorganized and shit their UX is. It’s terrible mindset over there.
great video! Thanks for this. Seems like its a lot simpler than having to understand how the tokenizer and architecture works and try to get your training dataset in that format. That being said, once tuned can the fine tuned model weights be downloaded to your local machine
Hi Thanks for letting me know - here is the JSON. below is a sample of one line items I used in my JSONL file. {"input_text": "question: What is Tyrone's view on the future of AI in insurance? context: AI is possibly the most exciting technology to ever have been created in my view, a solution that allows access to model and information at unprecedented rates","output_text": "The Gen AI future is looking awesome, so excited and can't wait to help more business adopt it"}
It took vertex AI to fine tune the data 2 hours 12 minutes on my side just for 10 examples while i was testing. Is there a way to speed this up? My original data has 14467 examples. It would take ages....
Hi, Thanks for watching and appreciate the comment. In Vertex AI, I have seen a way to allocate a more powerful server/TPUS's to the environment . Alternatively you may try the Google colab version which allows you to select the processing power you would like. I have only very briefly experimented with runtimes and different CPU configurations.
thank you for sharing.but Why am I getting an error in my app? google.api_core.exceptions.PermissionDenied: 403 You do not have permission to access tuned model tunedModels/finalemotions-xxxxin9vxxxu2lxxxvuf0. but I have API from AI Studio to access my tuning model to my app.
thanks for this video but Why am I getting an error in my app? google.api_core.exceptions.PermissionDenied: 403 You do not have permission to access tuned model tunedModels/finalemotions-xxxxin9vxxxu2lxxxvuf0. but I have API from AI Studio to access my tuning model to my app.
i have 500 examples and i did exactly the same way and Now it is more than 2hrs and it is still showing 'in progress' i can't even see how much completed or when it will complete, it's so confusing anyone please explain me
Hi Fine tuning models does take time, it is essentially taking a LLM and customising it. If you are looking for more visibility and control consider using the Vertex AI fine tuning methods. I experimented with a few hundred and it took 7-8 hours. It all depends on the TPU/CPU and servers allocated.
Hello.. Thank you for the video.. I have a question.. for inputs i have examples questions and 5 other input parameters and or course and output. How can i integrated there parameters into the trainin data set? Which format should i use? Thanks.
Hi thanks for watching, I would check the documentation as a first port of call. There may be some experimentation here. So keep the number of samples low to test it first.
I will definitely implement it! PS: You sound really experienced in this stuff, do you know of anyone that's interested in a educational/inspiring thumbnail? I am a graphic designer and planning on perfectly recreating the ideas of content creators... So I'm planning on asking the right questions, matching the brand, studying the viewers, matching the personality... and then using the right techniques (sometimes I'd like to jump out of the comfort-zone and learn new stuff) so that I can keep adjusting till they're satisfied with their ideal thumbnail.
hey cool idea, I am sure there are some people out there that would need help with the thumbnails. I am still quite new to RU-vid and "marketing" so you can use me as a test :)
Hi Yes, this is possible, i have seen previews of the Imagen (still not GA for all locations in the world). And then using the actions you can link the inputs to the Gemini text or Imaging models. I will do some more experimentation when i get a chance.
thank you for the video,i created the model as demoed in video but it shows 403 for my api key i followed all the steps exactly though google.api_core.exceptions.PermissionDenied: 403 You do not have permission to access tuned model
I really love your videos. You make it look so easy. I've been trying to combine two of your tutorials to make RAG application using vertex AI builder, and then integrate it to flutter flow. It would be absolutely amazing if you made a video on this.
Hi Thanks for watching, the use case for refining a LLM is to provide it with context for your data/environment. For instance if your marketing material is based on a certain style or PR questions and answers follow specific format, or if you want the model to behave like your organisation. Or if you are a specialised insurance/finance with use cases that you have as knowledge banks then using your business IP you can train the LLM further to be more accurate.
Hey, The Video was Great, I have a question. I have used RAG with a HuggingFaceLLM to make a Chatbot to query a few research papers. Now I want to fine tune this model and replace it in my code How should I generate Q/A prompts from the research papers to fine-tune the model such that the results are factual?
Hi that is a good question, I would take the research papers and look at what are question and answer type of content. *hint use the Gemini Pro with the 1.5 million tokens to pass in a PDF the paper and get it to suggest good question and answers for you. then fine tune from there.
I have one question im fine tune with my own QA data but when i want to CURL the api with the code of "Get code" a recieve a 403 error im not authorize to use that model but is my unique api key and my own fine tune model idu what is happening :/
Hi Thanks for watching. Because this is your private model and no one can use it. You will need to enable the authorisation of the API for your environment - to do this you can see documentation here: ai.google.dev/docs/model_tuning_guidance - check the the OAuth setup tutorial. This should help you get your authorisation working to access your private model.
@@LevelUp_Plus Even though i did the OAuth and it was successful still I am getting the same error "google.api_core.exceptions.PermissionDenied: 403 You do not have permission to access tuned model tunedModels/personal-saz4e2skuvn." please help me
i've got an error like this Failed to process dataset for gemini-1.0-pro-002: Dataset validation failed: {"missing_messages_list": while tuning model. What am i supposed to do?
Hi thanks for watching, I found the biggest reason for the failures was the JSONL file. Check the format of the file and ensure the format and end lines etc are correct.
Thanks for sharing. I'll share it with a good friend of mine in the insurance industry. This can definitely help her get promoted 😊, knowing how to automate working. Cheers!
Thanks for your very instructive video ❤❤❤. FlutterFlow sounds easy to use. I'm currently learning GLIDE to build apps for my daily routines. When I'm done, I'll try the same in FlutterFlow. Until your next video😊
Hi, thanks for watching. For the pricing, you will be charged storage costs of the data and your information you uploaded or indexed. For this example it was low cents per day. The second part of the costing is per query your chatbot does, you can check latest pricing from google here: cloud.google.com/generative-ai-app-builder/pricing. At the moment there is free credits applied, so be sure to use those.
hi sir, before I want to thank you for your content. Coincidentally, my final college assignment was to make a chat bot like this, would you like to make this chat bot so it can be entered into a database? the elements I use are python, vs code and gemini Ai
Hi thanks for watching, you can have a look at my videos on How to build Gemini chatbot in FlutterFlow OR StreamLit OR HTML for chatGPT - all 3 are simple follow alongs. FlutterFlow Chatbot: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-OHIgdwDwM4k.html StreamLit ChatBot: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-lkASdU_wdnc.html HTML: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-7wOlL1QmQFo.html
