Pydantic

oomarhelwe

class Clothing(BaseModel):
"""Data model for clothing items"""
name: str
product_id: int
price: float
class ClothingList(BaseModel):
"""A list of clothing items for the model to use"""
cloths: List[Clothing]
using openai agent :=== Calling Function ===
Calling function: inventory_query_engine_tool with args: {"input": "lightweight, short-sleeved linen shirt in a pastel color, such as a soft blue or mint green, for men"}
Got output: Error: 1 validation error for ClothingList
cloths
Field required [type=missing, input_value={}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.7/v/missing
========================

=== Calling Function ===
Calling function: inventory_query_engine_tool with args: {"input": "tailored chino shorts in a neutral color like beige or light grey for men"}
Got output: Error: 1 validation error for ClothingList
cloths
Field required [type=missing, input_value={}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.7/v/missing
========================

=== Calling Function ===
Calling function: inventory_query_engine_tool with args: {"input": "comfortable loafers or espadrilles for men"}
Got output: Error: 1 validation error for ClothingList
cloths
Field required [type=missing, input_value={}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.7/v/missing
========================

19 comments

WWhiteFang_Jr

Your llm wasn't able to parse info from the given text correctly

oomarhelwe

how can i fix it?

oomarhelwe

im using 4o

oomarhelwe

does that mean for the input for exemple "lightweight, short-sleeved linen shirt in a pastel color, such as a soft blue or mint green, for men" , the inventory query engine didnt get any matching ? or thats not the case?

LLogan M

Yea this means the LLM wasn't able to predict the output class properly

LLogan M

try adding better descriptions maybe

Plain Text

from pydantic.v1 import BaseModel, Field

class Clothing(BaseModel):
    """Data model for clothing items"""
    name: str = Field(description="Name of the clothing.")
    product_id: int = Field(description="Product ID of the clothing.")
    price: float = Field(description="Price of the clothing.")

LLogan M

But I also don't really know how you supplied these pydantic items to your agent/query engine

oomarhelwe

The vector store that the query engine is being used on should contain only name , id and price embeddings ?

oomarhelwe

this is the text in the vector store:
product_id: 1 name: Crocs Classic Lined Slingback Clogs description: description: Everybody loves the comfort of the Crocs Classic Clog — and now there’s a toasty lined version to keep the feeling going all season. The soft, fuzzy liner adds to the cushion and comfort, indoors or out. Great as a slipper, yet capable to run errands, too. Croslite™ foam construction keeps them light and easy to wear. The pivoting heel strap gives you a secure fit, or push it forward to just step in and go. The legendary Classic Clog, now with a warm, fuzzy liner Incredibly light and easy to wear Pivoting heel straps for a more secure fit Great indoors or out Customizable with Jibbitz™ charms Dual Crocs Comfort™: Blissfully supportive. Soft. Cradling comfort. color: White brand: Crocs category: Footwear product_type: flats

oomarhelwe

this is the pydantic model:
class Clothing(BaseModel):
"""Data model for clothing items"""
name: str = Field(description="Name of the clothing.")
product_id: int = Field(description="Product ID of the clothing.")

class ClothingList(BaseModel):
"""A list of clothing items for the model to use"""

cloths: List[Clothing]

oomarhelwe

using query engine:res=inventory_query_engine.query("Crocs Classic Lined Slingback Clogs")
print(res)

oomarhelwe

ValidationError: 1 validation error for ClothingList
cloths
Field required [type=missing, input_value={}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.7/v/missing

oomarhelwe

how there is a field required>

oomarhelwe

i used now gpt 4 and 3.5 turbo, the error is gone but im getting empty dictionnary now even though im writing the same name of products existing in the vector store

oomarhelwe

I think there is a problem with the pydantic parsing,I Used the llama index tutorial about Paul Graham with biography basemodel and im getting empty dictionnary

oomarhelwe

i remove the base model its working properly

oomarhelwe

The output parser is working only whith completion program

oomarhelwe

with any type of query engine its giving empty dictionnary

oomarhelwe

@WhiteFang_Jr

Add a reply

Find answers from the community

Pydantic