LangChain Tutorial (Python) #3: Output Parsers (String, List, JSON)

#openai #langchain

In this video we will add Output Parsers to convert the output from the LLM into String, Arrays and JSON schemas.

☕ Buy me a coffee:

📑 Useful Links:

💬 Chat with Like-Minded Individuals on Discord:

🧠 I can build your chatbots for you!

🕒 TIMESTAMPS:
00:00 - Introduction to Output Parsers
01:08 - Getting things ready
02:56 - String Output Parser
04:08 - Adding Parser to Chain
05:03 - Comma Seperated Output Parser
07:00 - JSON Output Parser
11:54 - Structure with List

[rrioclkls]

your videos have brought me back from the brink of insanity, thank you.

[alhabib_o]

I just wanted to say thank you for your incredible tutorials. I’m learning a lot and loving every minute of it!

[daltheman]

As a tech teacher for over 20 years, I have to say that your teaching is excellent. You are beyond just being a programmer. Thank you and keep teaching us!

[ahassan]

You are wonderful man. You have EXTRAORDINARY way in delivering and explaining information. I really get great benefit from your channel.
Thank you so much for your great efforts and sharing such important information.👏👏

[lindseyfisher]

These videos are incredibly helpful! I've watched tons of tutorials and yours just make the concepts so clear and easy to follow! Subscribed and look forward to more

[daedae]

Such a great course, you have an amazing teaching style. Also, for others using langchain 0.3 now import pydantic like so "from pydantic import BaseModel, Field". V1 is deprecated.

[ishandindorkar]

Thank you for creating this video. It helped me a lot.

[Matt-hpcl]

Excellent tutorial, so clearly explained, thank you. I needed the JSON parser.

[AssassinUK]

This and your previous video have been great! But I started following you based on your Flowise videos and now I use Flowise a lot for production work as well. Are you going to make a Flowise version of your videos based on these videos at all? Don't forget your Flowise followers 😉

[fatiga]

Hi, great video!
I was wondering, would it be fine if we just got the string from the content of the AI Message Object instead of using a parser?
I noticed everyone uses parsers for this but I'm not sure why yet.
Thanks!

[rogerwmwong]

I am trying to follow your tutorial. When I am writing the statement "parser = JsonOutputPaser(pydantic_object=Person)", there is no "pydantic_object" although I can find "pydantic" and "pydantic_core", should I use "pydantic" instead?

[iammadhankumar]

Hi, which model should we use for code migration? For example, migrating a Java servlet controller to a Spring Boot controller.

[alokbhatt]

I am trying to extract 3rd field as address from my prompt apart from name and age. But I am not getting any output why? Is it because we can only output two fields from the json parser?

[overprismatic-rvci]

Why your terminal has such cool colors? Mine is all white. Also my code has different colors...

[CreatewwwPl]

Leo, You should import `pydantic.v1` instead of `langchain_core.pydentic_v1`.