This video is as usual very informative. Regarding the setup - I vote for the old setup. In this new setup, around 30%-40% screen is getting covered by your movement and we are not able to see the screen entirely which was not the case in earlier setup. If you want to adept to this new setup then keep some space between the code screen and yourself.
You were wrong about Refine method It doesn't use chunk 1, 12, 123, 1234.n It takes the summary of the first chunk and combines with the next chunk saying "till now this is the summary, can we refine it by this additional text"
Dear Krish, thank you for this video. for the last part summary chunks, how can we indicated the length of summary & the selected language for the final output ?
Great stuff Krish, love the new setup. Btw, can you do a text summarization where the conversation is taking place between a customer and support agent.
Hi Krish Sir, can we do some question answer based on a repository as well, like i gave the full github repo and asking the question like what is the data flow, how many flow present in this repository. what is the agenda of the repo, etc... is yes how we can do it any guidance ?
Hi Krish sir, Is it possible to generate summary by uploading 50 page document at one go by using open source LLM models / free versions ? If yes then can you pls guide
I've read many comments from the internet that langchain is not a production grade framework. Meaning it is not advisable to use langchain in the production. Is it true Krish??
Hi Krish, I have doubts about Refine chain type if we are combining chunks for every chunk for the last chunk it will get the whole documentation again right then why does the need to divide the document into chunks
better create a requirements.txt file in the same directory and put all the required packages and run it later pip install -r requirements.txt. It will solve the issue
I liked your old setup earlier where there was the visibility of the code and we can get the walk through.. this is just talking so I prefer the previous setup even if it was just copy paste from any different notebook. The previous setup would allow us to see where the code could go wrong. This seems more like a lecture format which everyone is doing. TLDR; liked your old setup. Also you look uncomfortable!!
Just trying new things...I hope you like it :)
Yes sir i liked it
You are amazing, thank you for this it was more engaging, thank you very much
Much needed
Very nice new one but your old setup will forever gold ❤ loved that
may I know which you are mentioning ?. I am new to this channel
Krish and his tutorials rocking as always. And Krish's makeover is an added bonus
Loved this new setup feel more grasping
This video is as usual very informative. Regarding the setup - I vote for the old setup. In this new setup, around 30%-40% screen is getting covered by your movement and we are not able to see the screen entirely which was not the case in earlier setup. If you want to adept to this new setup then keep some space between the code screen and yourself.
may I know which you are mentioning ?. I am new to this channel
love the new video format Krish!
you pretty much go though the langchain summarize document page . but have to admit , this is faster and on to the point . So Thank you!
I dont know how to thank you Sir! You save me every single time!!!
You were wrong about Refine method
It doesn't use chunk 1, 12, 123, 1234.n
It takes the summary of the first chunk and combines with the next chunk saying "till now this is the summary, can we refine it by this additional text"
Super Helpful ! Thanks Krish❣
Love this setup sir! Amazing🔥🔥🔥🔥🔥🔥🔥🔥
Very Good Setup . I am waiting for real business dataset Project
We will learn from u but how will u learn these all new technologies krish sir
omg! love the setup
Do more video like this
nice enviroment like stack market back ground screen graph 🙂
Great video. Thankyou for sharing. Btw, I liked yourearlier setup more.
Super helpful thank you
informative...nice explaination
How can we evaluate those generated summary? Any metrics or framework, if any one can suggest
Hi Krish , can you pls make a video related to summarization of conversational text data using prompt engineering ?
Dear Krish, thank you for this video. for the last part summary chunks, how can we indicated the length of summary & the selected language for the final output ?
Waiting for End To End project with good UI sir....
Great video just a suggestion sir whatever you mention to go to your referred video,plz include that link in your detail part.
Thanks
Very useful and nice sir
can i use the map reduce technique to generate french summary?
Great stuff Krish, love the new setup. Btw, can you do a text summarization where the conversation is taking place between a customer and support agent.
Sir please teach azure with ML.
Can we do analysis of incident dump and find out opportunities for improvements
Is it possible to get deterministic summary from these techniques??
how did you validate the result? What metric did you use?
❤❤❤
Hi Krish Sir,
can we do some question answer based on a repository as well, like i gave the full github repo and asking the question like what is the data flow, how many flow present in this repository. what is the agenda of the repo, etc...
is yes how we can do it any guidance ?
How to increase the length of summary in StuffDocumentChain Text summarization
Is it possible to control the lengh of the summary? like summarize in about 200 words or summarize in 50 words
I think the old view was good because here things are getting covered behind your movement
same
No
Hi Krish sir,
Is it possible to generate summary by uploading 50 page document at one go by using open source LLM models / free versions ?
If yes then can you pls guide
I've read many comments from the internet that langchain is not a production grade framework. Meaning it is not advisable to use langchain in the production. Is it true Krish??
Hi can you share gen ai playlist
thank you so much
Sir please teach block chain
Hi Krish, I have doubts about Refine chain type if we are combining chunks for every chunk for the last chunk it will get the whole documentation again right then why does the need to divide the document into chunks
Refine is using the summary from chunk 1 and combine with chunk 2 and push to LLM.
So assume I have 30 000 customer reviews for one product. How can I ask the LLM to summarize all the reviews into one master review?
Sir its a very informative video but do make sure to hide your api key aur maybe after the video delete that ai api key
thank you!
Namaste
except ImportError:
--> 302 raise ImportError(
303 "Could not import openai python package. "
304 "Please install it with `pip install openai`." getting this error sir
!pip install openai
(and then in next cell)
import openai
better create a requirements.txt file in the same directory and put all the required packages and run it later pip install -r requirements.txt. It will solve the issue
The token size is 8192 precisely
Modi ka naam sunke Mood off ho gaya..puri duniya me ye Desh dhrohi ka speech hi mila ?
😂
I liked your old setup earlier where there was the visibility of the code and we can get the walk through.. this is just talking so I prefer the previous setup even if it was just copy paste from any different notebook. The previous setup would allow us to see where the code could go wrong. This seems more like a lecture format which everyone is doing. TLDR; liked your old setup. Also you look uncomfortable!!