[Day 81] RAG from scratch - chunking is very important!

 Hello :)
Today is Day 81!


A quick summary of today:
  • Continued with my custom RAG from scratch project - extracting knowledge from a bank's terms and conditions (github repo)


So, yesterday my main struggle was reading tables. I remembered gemini can read pictures so I gave the below to gemini to try to read and give me the text. 
The output, not very good. Gemini could not read this table very well, and such tables were common in yesterday's PDF
Yes, this is an image, but even if a powerful gemini LLM could not read this table and output it as text, for now at least, I gave up on this particular PDF, and looked for one with a bit more straigh-forward text and less tables. 

The newly chosen pdf is here: on github

Now... given an allegedly more simple pdf, I used the code as it was to get outputs. But the results were just... really bad. Most times even though the scores were high, and the top-1 included the exact answer, the output was  ~'the context does not include {query} information'.

I decided to go back to the start, where I preprocess the pdf, and explore more chunking mechanisms. I found this youtube video exploring different chunking methods.

It covered:

  1. Character split (which I was using)
  2. Recursive character split
  3. Document spcific splitting
  4. Semantic splitting
  5. Agentic splitting
The youtuber mentioned that the default method he uses is the 2nd one - recursive character split, so I decided to move from 1. to 2., and explore how recursive splitting improves my process. 

This is the final text_splitter I went with:
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1500, chunk_overlap=0) (tbh, I am writing this blog now and 1500 might be an overkill)

I played around a lot with chunk_size and chunk_overlap - to see the effects, and I don't think I mastered knowing the impact yet. I am sure 1500 chunk_size is even that good. Frankly, in the preprocessing there are other components like splitting, regex-ing characters - maybe there are better ways to do these. It will take trial and error to figure this. 

The query list I used to see results is:

Base prompt is like this

I am yet to explore different prompting strategies. So far only concentrated on the preprocessing step. 

After settling on chunk_size 1500 and chunk_overlap 0, I decided to streamline the process a bit, so I created executable preprocess_pdf.py and rag.py files - can be seen in the repo.

As for results, they are incosistent... 

Example is this 'What is meant by Business Day?' query.
The top-1 context includes the answer, but some times I get the answer, sometimes I do not.

I shared my code on reddit and on some ML discord server, to look for advice. But tomorrow I think I shall concentrate on finding a better chunking method. 



That is all for today!

See you tomorrow :)

Popular posts from this blog

[Day 198] Transactions Data Streaming Pipeline Porject [v1 completed]

Image
 Hello :) Today is Day 198! A quick summary of today: data streaming pipeline project [v1 done] Here is a link to the project's repo. Well ... I did not know I can do it in a day (~14 hours) after yesterday's issues but here we are. Turns out in order to insert the full (~70 variables with nested/list structure), I need the proper pyspark schema. And yesterday I did not have that and that is why when I was reading data in the kafka producer I was getting NULL in the columns - my schema was wrong. Well today I not only fixed the schema for the 4 variables I had yesterday, but included *all* the variables that come from the Stripe API ~ 70 (for completeness).  When I run docker-compose, the data streams and is input into the postgres db (and is still running). Unfortunately, the free Stripe API for creating realistic transactions has a limit of 25, so every 3 seconds, 25 new transactions are sent to the db. It has been running half the day (since I got that set up) and as I am w...
Read more

[미리 공부] 기초 통계 복습 (Day 1는 1월2일)

Image
모험을 시작하기 전에 기초 지식을 복습하고자 했다. 오늘 SPSS 대안 프로그램을 찾아보려고 해서 JASP에 대해 알게 되었다. 유용한 프로그램인 것 같아서 선회귀와 기술통계를 내려고 했는데 재미있었다. 그런데 JASP에 대해 더 알기 전에 '기초 통계 지식을 좀 복습을 하고자 하면 좋을 것 같아'란 생각을 들었다. 다행히, Coursera에서 Stanford University의 Guenther Walther 교수님께서 진행된  Introduction to Statistics  무료 강좌가 있다.  좀 부족한 부분은 다양한 검정통계 하는 거고 (F test, t-test, chi-square 등) 이제 JASP 아니면 다른 통계 프로그램 사용하게 되어도 이런 부분을 좀 더 자세히 집중하여 공부하면 된다.  특히 homoscedasticity 및 heteroscedasticity 개념을 기억에 남았다. Homoscedasticity (선): Definition: In a homoscedastic dataset, the variance of the errors (residuals) is constant across all levels of the independent variable(s). In simpler terms, the spread of the residuals is the same throughout the range of predictor values. Heteroscedasticity (악): Definition: Heteroscedasticity occurs when the variance of the errors is not constant across all levels of the independent variable(s). In other words, the spread of residuals changes as the values of the independe...
Read more

[Day 61] Stanford CS224N (NLP with DL): Machine translation, seq2seq + a side CDCGAN mini project

Image
 Hello :) Today is Day 61! A quick summary of today: Covered Lecture 7 : machine translation, seq2seq, attention from Stanford CS224N Tried to make a conditional DCGAN to generate MNIST numbers ( colab ) ( kaggle ) I will first cover the GAN story (then will share my notes from the lecture) So... while watching and taking notes today, I started thinking, what if I can use my notes as data to a model and afterwards, when I want, I can give it raw string text and it will output text in the format of my notes (with my handwriting). Well I started looking around and actually the first model architecture that came to my mind was the GAN (specifically conditional GAN) - I remembered there was a GAN architecture that alongside the pictures, we can give it the labels, and then on-demand generate. In retrospect, there are of course others, but I decided to go with GAN.  For maybe 2 hours I busted my head trying to make a simple model with the EMNIST dataset (english characters), and I ...
Read more