PLease upload on the Fine tuning of the Llama 2 models (7B, 13B et), especially regarding how to create a data set for a chatbot to model fine tuning and model inference
Hello. I've been thinking about ways to improve response speed in this example, and I came across fine-tuning the model using autotrain. I was wondering if fine-tuning the model with GPU support from platforms like Kaggle or Colab could be a viable option for achieving faster responses. I think it would be really interesting if you could explore this topic in one of your videos and show us how to fine-tune a model using GPU support (this same example). It would be great to see the process in action and learn more about the benefits and considerations involved in fine-tuning models for faster response times. Looking forward to your insights on this topic! Keep up the great work on your channel. Jorge.
really very informative video. Thank you dear for your efforts and informative video. I appreciate your effort and content dear. Keep it up. Once again thank you ❤❤❤
Which sentence transformer are we using exactly? And why not any top model from mteb ? What are the drawbacks? Would be helpful if you can clarify. Great video tho❤
It seems like there's a mismatch that I can't figure out. The model has dimensions 384 and the datastore 768. So it won't update any dimensions: RuntimeError: Embedding dimensions of the model (384) don't match the embedding dimensions of the document store (768). Initiate WeaviateDocumentStore again with arg embedding_dim=384. What am I missing?
Hi. I am building a chatbot with my pdf document. I created an endpoint in Fastapi. If i send 1 request the answer is really fast, but if i try to send 100 requests in the same time the response time is so much bigger. How can i run those requests paralell using gpu?
thanks for the informative video,can you upload a same tutriol video for different types of document summarization implementing RAG and open source LLM not OpenAI asap...it would be greatful need help in that and not found any related video for summarization .
help !! Tried everything i know to resolve this problem (I have cloned your git repo) Error: ImportError: cannot import name 'Document' from 'haystack' (E:\ATULYA\Haystack\myenv\Lib\site-packages\haystack\__init__.py)
I had same problem, fixed it by updating requirements.txt. Remove "farm-haystack" and keep only "farm-haystack[weaviate]". You need to pip uninstall farm-haystack. It seems there is some conflict, when both libs are installed and some classes are not visible
thanks again this is my exact use case here in DC! Just testing different opensource stacks there is a great langchain interview w Weviate discussing the technique of using a summary vector in each prompt to improve LLM responses. How would you improve the response time?
Hi , thank you for the tutorial , when i'm install the requirements.txt , I got this error message : ERROR: Could not find a version that satisfies the requirement torch (from versions: none) ERROR: No matching distribution found for torch . Can you please advise and help? Thank you
Thank you very much, you are fantastic, you could be a teacher your natural way of talking is very captivating !!!!, what do you think about PGVector ?
i try finishing this pending video, and i have a question. today i see your video of jina v2. in this video you says about model of incrustation in huggingface that name is multilingüe. Is possible to fusion? you can make a video for best implementations cases that jina 2 or ideas for other cases of language spanish or catalan jeje :P thanks for your content! is increible!
i am facing an issue with embedding dimension. this is the error: Embedding dimensions of the model (384) don't match the embedding dimensions of the document store (768). Initiate WeaviateDocumentStore again with arg embedding_dim=384
Hey bro! very nice video, what are the system specs for running this project? Is 8gb RAM enough, to produce results if we use Quantised Bloke mistral model
Hi sir Am making a pdf questions answers chatbot but the issues ia my dataset is to dirty and aslo very small So what's u r suggestion without changing the dataset how can i achieve the average accuracy I know its a silly question but ...
Oh Man, you rock! it seems like you read your followers' minds. Thank you so much for this great video.
Yeah Non Invasive models in brain has been implanted 😂.. thanks btw
PLease upload on the Fine tuning of the Llama 2 models (7B, 13B et), especially regarding how to create a data set for a chatbot to model fine tuning and model inference
Absolutely great quality content. Please continue the great work, These videos are extremely useful.
Thanks Shivam... sure more on the cards.
Haystack the same time @@AIAnytime
Hello.
I've been thinking about ways to improve response speed in this example, and I came across fine-tuning the model using autotrain. I was wondering if fine-tuning the model with GPU support from platforms like Kaggle or Colab could be a viable option for achieving faster responses.
I think it would be really interesting if you could explore this topic in one of your videos and show us how to fine-tune a model using GPU support (this same example). It would be great to see the process in action and learn more about the benefits and considerations involved in fine-tuning models for faster response times.
Looking forward to your insights on this topic! Keep up the great work on your channel.
Jorge.
This should be addressed
This is just next level
Sir, can you please show how to run it in Mac M1 system as well
Yes please sir, it would be helpful for me also
Thank you for doing a video on this!
Glad it was helpful!
really very informative video. Thank you dear for your efforts and informative video. I appreciate your effort and content dear. Keep it up. Once again thank you ❤❤❤
It's my pleasure
Thanks, great tutorial
Glad it was helpful!
Hi,
The embedding dimension specified is 768 and you are using a 384 model. It throws error, can you help?
AI now?! Let's go!
Let's go
At 1:40 you said you are going to use "Vector Database" not "Vector Store". Can somebody please tell me the difference between them.
Hi! Duda from Weaviate here. They are the same. Langchain calls the. Vector Store, but they are the same :)
hi i really love you video weldone but am finding it hard to install haystack on my system it saying haystack.nodes is not found.
please any help
Which sentence transformer are we using exactly? And why not any top model from mteb ? What are the drawbacks? Would be helpful if you can clarify.
Great video tho❤
THIS GITHUB REPO IS NOT UPDATED AND IS GIVING MULTIPLE ISSUES!
It seems like there's a mismatch that I can't figure out. The model has dimensions 384 and the datastore 768. So it won't update any dimensions:
RuntimeError: Embedding dimensions of the model (384) don't match the embedding dimensions of the document store (768). Initiate WeaviateDocumentStore again with arg embedding_dim=384.
What am I missing?
Can you do it without Docker?
Since Docker needs a commercial licence for bigger companies..
thankyou sir sir i want use sqldatabase inplace of pdf and etc
so please make video for mysql
This git repository has not been updated. Can you upload models, template files/folders
Hi. I am building a chatbot with my pdf document. I created an endpoint in Fastapi. If i send 1 request the answer is really fast, but if i try to send 100 requests in the same time the response time is so much bigger. How can i run those requests paralell using gpu?
+1
thanks for the informative video,can you upload a same tutriol video for different types of document summarization implementing RAG and open source LLM not OpenAI asap...it would be greatful need help in that and not found any related video for summarization .
help !!
Tried everything i know to resolve this problem (I have cloned your git repo)
Error: ImportError: cannot import name 'Document' from 'haystack' (E:\ATULYA\Haystack\myenv\Lib\site-packages\haystack\__init__.py)
im getting same error on mac python 3.11
is there any particular version of python or haystack we have to use?
did you get the solution for this error??
I had same problem, fixed it by updating requirements.txt. Remove "farm-haystack" and keep only "farm-haystack[weaviate]". You need to pip uninstall farm-haystack. It seems there is some conflict, when both libs are installed and some classes are not visible
thanks again this is my exact use case here in DC! Just testing different opensource stacks there is a great langchain interview w Weviate discussing the technique of using a summary vector in each prompt to improve LLM responses. How would you improve the response time?
Yeah that's a very good idea. Those can be used in memory as well. Give it a try.
Let me know what you build. Thanks
Hi , thank you for the tutorial , when i'm install the requirements.txt , I got this error message : ERROR: Could not find a version that satisfies the requirement torch (from versions: none)
ERROR: No matching distribution found for torch . Can you please advise and help? Thank you
awesome thank you!
You're welcome!
excellent !
Many thanks!
Thank you very much, you are fantastic, you could be a teacher your natural way of talking is very captivating !!!!, what do you think about PGVector ?
Thank you! 😃 yes You can try Supabase. It has a PG extension i guess.
i try finishing this pending video, and i have a question. today i see your video of jina v2. in this video you says about model of incrustation in huggingface that name is multilingüe. Is possible to fusion? you can make a video for best implementations cases that jina 2 or ideas for other cases of language spanish or catalan jeje :P thanks for your content! is increible!
Great suggestion. Will create one soon...
❤❤Viagra
Bro when’s the “deployment on aws” video coming out ?. It’s been more than a month since the 2 nd part released
i am facing an issue with embedding dimension. this is the error: Embedding dimensions of the model (384) don't match the embedding dimensions of the document store (768). Initiate WeaviateDocumentStore again with arg embedding_dim=384
I had same issue try delete the docker image and again rerun everything starting from creating docker image.
Can please explain where you download the two models you have the mistral and the llama
Huggingface repo.
can you explain why this all is good thing?
people talk about it as awesome stuff. but why?
bhai thoda mike ki quality sai krr lo yaar
aawaz chubh rai h kasam se
Baaki great knowledge
keep up the good work
I just love how you pronounce massine 😂
Lol 😂
Hey bro! very nice video, what are the system specs for running this project? Is 8gb RAM enough, to produce results if we use Quantised Bloke mistral model
Hi sir
Am making a pdf questions answers chatbot but the issues ia my dataset is to dirty and aslo very small
So what's u r suggestion without changing the dataset how can i achieve the average accuracy
I know its a silly question but ...
Hello! can i add multiple pdfs and can it read images in pdfs?
Dependency management in Python is a pain. Which exact package version to install? Nobody knows 😂
Use poetry for managing the versions.
same! ImportError: cannot import name 'send_event' from 'haystack.telemetry' (/usr/local/lib/python3.8/site-packages/haystack/telemetry/__init__.py)
check your mic or audio settings. totally over-driving the sound 🤗
Thanks for the tip!
Hey did you try fine-tuning mistral on custom domain data? Can you recommended GPU for fine tuning 7B models such as Mistral, Llama etc..?
Take a rental GPU on Runpod or A100 or v100 Google Colab with high RAM.
Please provide the link for downloading the quantized Mistral model
Here you go: huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF
_fire_
Weaviate is better or Pinecone?
Pinecone premium is better. Weaviate is also good as open sourced.