I followed your content 2 hours before the exam. Passed 2 min ago, thought you deserve to know. I did get few q from your slides and few from the other videos I followed. Your explaination really helped in understanding the difference for the last minute revision. I would highly recommend. my next one is the OCI security - do you think you can create on for that? Thanks again.
21 is C -- Conversational Image Memory is the odd one out in terms of OCI Gen AI's offerings. The other three options are all closely related to text-based conversational AI, while Conversational Image Memory would require additional capabilities and features not typically associated with OCI Gen AI.
Thanks for this great video! Q21 is not a is c: there is no LangChain memory type as ConversationImageMemory but there is indeed a ConversationTokenBufferMemory...btw, is Conversation not Conversational...
25 is A in my opinion. If I had to choose between Diffusion Model and RAG, I'd say that Diffusion Model is more suitable for image-to-text tasks, especially when it comes to generating descriptive text from images.
I took the OCI generative AI Exam yesterday and passed with 90%. Thanks Preet, your explanations helped me a lot in clearing this Exam with ease. I really appreciate.
Q 21, correct answer should be C: ConversationImageMemory All these 5 are commonly used: ConversationBufferMemory ConversationBufferWindowMemory. ConversationTokenBufferMemory ConversationSummaryMemory ConversationSummaryBufferMemory except ConversationImageMemory
Yes, LLMs Can Describe Images LLMs (Large Language Models) can indeed describe images. However, they require a specific type of model known as a multimodal LLM to do this effectively. These models are trained on both text and image data, allowing them to understand and interpret visual information.
Hi, thankyou for detailed explanations on the answers, really helped me complete the course and successfully passed it. I'm struggling to download the certification, can you please guide me if possible? Or does it take time to generate the certificate? It already shows that I've passed the exam.
thanks for making this video, great effort, just minor things, While I was watching I think for Question 21, the answer should be Conventional Image Memory is odd, and In Question 31, you said option B is correct, which is right, though on the Screen It shows C is correct.
I also think that the option C is more relevant for 3rd question, as I don't see anything about reducing the number of parameters anywhere. For reference, here is the content from the official Oracle University videos: T-Few Fine-tuning: Traditionally, Vanilla fune-tuning involves updating the weights of all (most) the layers in the model, requiring longer training time and higher serving (inference) costs. T-Few fine-tuning selectively updates only a fraction of the model's weights. It is an additive Few-Shot Parameter Efficient Fine tuning (PEFT) technique that inserts additional layers, comprising .01% of the baseline model's size. The weight updates are localized to the T-Few layers during the fine-tuning process. Isolation the weight updates to these T-Few layers significantly reduces the overall training time and cost compared to updating all layers. T-Few fine-tuning process: It begins by utilizing the initial weights of the base model and an annotated training dataset. Annotated data comprises of input-output pairs employed in supervised training. Supplementary set of model weights is generated (~0.01% of the baseline model's size). Updates to the weights are confined to a specific group of transformer layers, (T-Few transformer layers), saving substantial training time and cost.
Doubt regarding the OCI Gen AI certification. Do I need to attend the exam within July 31st. Or register within the specified date and can attempt later? (for getting free)
Is attending a live session is necessary to get the certificate as i am not getting mail fir live session even after registering for the live session. Its showing me waitlisted in the oracle my learn website. What to do please help
I too got the same doubt. A quick check revealed some interesting answers. From OCI Generative AI documentation: An endpoint is a designated point on a dedicated AI cluster where a model can accept user requests and send back responses such as the model's generated text. You create an endpoint on a hosting dedicated AI cluster From Microsoft Azure documentation: An endpoint is a stable and durable URL that can be used to request or invoke a model. Both the options seem to be true initially, but after this reference, I think we need to select A instead of D
I followed your content 2 hours before the exam. Passed 2 min ago, thought you deserve to know.
I did get few q from your slides and few from the other videos I followed.
Your explaination really helped in understanding the difference for the last minute revision.
I would highly recommend.
my next one is the OCI security - do you think you can create on for that?
Thanks again.
i just passed my exam and this video was super helpful to revise before taking my exam! thank you so much 🙋♀️
21 is C -- Conversational Image Memory is the odd one out in terms of OCI Gen AI's offerings. The other three options are all closely related to text-based conversational AI, while Conversational Image Memory would require additional capabilities and features not typically associated with OCI Gen AI.
Gave my exam today and passed it, This video is indeed close to the actual question pattern. Thanks for the video!
Congratulations
Questions are exactly same?
Thanks for this great video! Q21 is not a is c: there is no LangChain memory type as ConversationImageMemory but there is indeed a ConversationTokenBufferMemory...btw, is Conversation not Conversational...
25 is A in my opinion. If I had to choose between Diffusion Model and RAG, I'd say that Diffusion Model is more suitable for image-to-text tasks, especially when it comes to generating descriptive text from images.
thanks
Thanks for creating and explaining the concepts.
I took the OCI generative AI Exam yesterday and passed with 90%. Thanks Preet, your explanations helped me a lot in clearing this Exam with ease. I really appreciate.
Hey hi
How was the exam did they ask the same questions only
Congratulations.. Great job.
@@tharuunmsame question
Great job! The answers explanations, both right and wrong ,really helped my understanding. I look forward to more videos from you on other topics.
Q. No. 21 conversationalImageMemory is odd since others are related to text based memory.
Q 21, correct answer should be C: ConversationImageMemory
All these 5 are commonly used:
ConversationBufferMemory
ConversationBufferWindowMemory.
ConversationTokenBufferMemory
ConversationSummaryMemory
ConversationSummaryBufferMemory
except ConversationImageMemory
I had passed! Thanks a lot for your video!! 😁
The 25th question's answer is Diffusion Model, as llms are not designed to analyse images.
LLM with fine tuning is possible!!
Yes, LLMs Can Describe Images
LLMs (Large Language Models) can indeed describe images. However, they require a specific type of model known as a multimodal LLM to do this effectively. These models are trained on both text and image data, allowing them to understand and interpret visual information.
for the 3 question option C can also be a correct option??
Thank you so much for your hard work. Very helpful..
Hi, thankyou for detailed explanations on the answers, really helped me complete the course and successfully passed it.
I'm struggling to download the certification, can you please guide me if possible?
Or does it take time to generate the certificate? It already shows that I've passed the exam.
Are the questions same as described in the video??
@@MUKESH-cl6km most of them could be same
@@karansonawane2392Thanks Karan, Could you please also confirm if the complete video is enough to Pass the exam or not??
thanks for making this video, great effort, just minor things, While I was watching I think for Question 21, the answer should be Conventional Image Memory is odd, and In Question 31, you said option B is correct, which is right, though on the Screen It shows C is correct.
Just to test our 1. seek and learn attitude and 2.listening skills😅
Q 31, option B is the correct answer.
Option b is correct it seems
for Question 31, option B is correct
This is enough to study from this video, will the questions will be asked same from.thjd video or different please help me
On question 40 I saw the same text for an and b, though what you explained for b explains why it was correct. Thanks for making the video, btw.
What's the right answer?
gr8 content. Q40 has answer option A and B are same.
How close are these questions to the examinations?
It will be great if you can provide those questions as the pdf.
Questions 40. A and B are duplicate options.
Can you please guide me total.how many questions asked
40 questions 90 minutes time
Great review with explanations thanks lady! :sunflower
For the third question the right answer is option C!!
I also think that the option C is more relevant for 3rd question, as I don't see anything about reducing the number of parameters anywhere.
For reference, here is the content from the official Oracle University videos:
T-Few Fine-tuning:
Traditionally, Vanilla fune-tuning involves updating the weights of all (most) the layers in the model, requiring longer training time and higher serving (inference) costs.
T-Few fine-tuning selectively updates only a fraction of the model's weights.
It is an additive Few-Shot Parameter Efficient Fine tuning (PEFT) technique that inserts additional layers, comprising .01% of the baseline model's size.
The weight updates are localized to the T-Few layers during the fine-tuning process.
Isolation the weight updates to these T-Few layers significantly reduces the overall training time and cost compared to updating all layers.
T-Few fine-tuning process:
It begins by utilizing the initial weights of the base model and an annotated training dataset.
Annotated data comprises of input-output pairs employed in supervised training.
Supplementary set of model weights is generated (~0.01% of the baseline model's size).
Updates to the weights are confined to a specific group of transformer layers, (T-Few transformer layers), saving substantial training time and cost.
I'm currently taking ai foundations anddd don't have the time to study for this. Will this video be enough to pass. I'm kinda new to ai
not this video alone but if you combine it with other videos and your own practice on dumps you should be good
Doubt regarding the OCI Gen AI certification. Do I need to attend the exam within July 31st. Or register within the specified date and can attempt later? (for getting free)
you need to attend this exam till 31 July to get free certification
Is attending a live session is necessary to get the certificate as i am not getting mail fir live session even after registering for the live session. Its showing me waitlisted in the oracle my learn website. What to do please help
Good job Preet:)
Thanks for sharing❤
For 31 B is the correct answer
You are right!
For 23, i feel option D is the right answer
I too got the same doubt. A quick check revealed some interesting answers.
From OCI Generative AI documentation:
An endpoint is a designated point on a dedicated AI cluster where a model can accept user requests and send back responses such as the model's generated text. You create an endpoint on a hosting dedicated AI cluster
From Microsoft Azure documentation:
An endpoint is a stable and durable URL that can be used to request or invoke a model.
Both the options seem to be true initially, but after this reference, I think we need to select A instead of D
@@SatishChandraAnnamaneni oh thank you so much Satish I'm taking exam tomorrow morning
Can you share your study notes please😊
I have covered almost everything on video.
Has anyone tried using Android phone as webcam and mic for the exam?
Nice
I have passed exam with 85%
@@kulwantkumarmba9741 Were the same questions repeated? If yes, all of them?
was the video helpful? How were the questions compared to the ones in video?
Has anyone tried using Android phone as webcam and mic for the exam?
you can't
@@EAEEOEAThank you for your response.