Create retriever.py
Browse files- retriever.py +87 -0
retriever.py
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from smolagents import Tool
|
| 2 |
+
from langchain_community.retrievers import BM25Retriever
|
| 3 |
+
from langchain.docstore.document import Document
|
| 4 |
+
|
| 5 |
+
class VegetableInfoRetrieverTool(Tool):
|
| 6 |
+
name = "vegetable_info_retriever"
|
| 7 |
+
description = "Retrieve information about vegetables based on their name or characteristics."
|
| 8 |
+
inputs = {
|
| 9 |
+
"query": {
|
| 10 |
+
"type": "string",
|
| 11 |
+
"description": "The name or description of the vegetable you want information about."
|
| 12 |
+
}
|
| 13 |
+
}
|
| 14 |
+
output_type = "string"
|
| 15 |
+
|
| 16 |
+
def __init__(self, docs):
|
| 17 |
+
self.is_initialized = False
|
| 18 |
+
self.retriever = BM25Retriever.from_documents(docs)
|
| 19 |
+
|
| 20 |
+
def forward(self, query: str):
|
| 21 |
+
results = self.retriever.get_relevant_documents(query)
|
| 22 |
+
if results:
|
| 23 |
+
return "\n\n".join([doc.page_content for doc in results[:3]])
|
| 24 |
+
else:
|
| 25 |
+
return "No matching vegetable information found."
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def load_vegetable_dataset():
|
| 29 |
+
# Manually define the vegetable data
|
| 30 |
+
vegetable_data = [
|
| 31 |
+
{
|
| 32 |
+
"name": "Sweet Potato",
|
| 33 |
+
"type": "root",
|
| 34 |
+
"description": "A starchy root vegetable with sweet flavor, rich in fiber and beta-carotene."
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"name": "Fresh Basil",
|
| 38 |
+
"type": "herb",
|
| 39 |
+
"description": "An aromatic green herb often used fresh in Mediterranean cuisine."
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"name": "Green Beans",
|
| 43 |
+
"type": "legume",
|
| 44 |
+
"description": "Long green pods with small seeds inside, eaten whole and rich in vitamins."
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"name": "Bell Pepper",
|
| 48 |
+
"type": "fruit vegetable",
|
| 49 |
+
"description": "Crunchy and colorful, bell peppers are technically fruits, but used as vegetables in cooking."
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"name": "Broccoli",
|
| 53 |
+
"type": "cruciferous",
|
| 54 |
+
"description": "Green vegetable with edible flowering heads and stalks, high in fiber and vitamin C."
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"name": "Celery",
|
| 58 |
+
"type": "stem",
|
| 59 |
+
"description": "Crisp green stalks often eaten raw or cooked, known for high water content and fiber."
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"name": "Zucchini",
|
| 63 |
+
"type": "squash",
|
| 64 |
+
"description": "Soft-skinned green vegetable, technically a fruit, often grilled or sautéed."
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"name": "Lettuce",
|
| 68 |
+
"type": "leaf",
|
| 69 |
+
"description": "Leafy green used in salads, comes in various varieties like romaine or butterhead."
|
| 70 |
+
},
|
| 71 |
+
]
|
| 72 |
+
|
| 73 |
+
# Convert each entry to a Document
|
| 74 |
+
docs = [
|
| 75 |
+
Document(
|
| 76 |
+
page_content="\n".join([
|
| 77 |
+
f"Name: {veg['name']}",
|
| 78 |
+
f"Type: {veg['type']}",
|
| 79 |
+
f"Description: {veg['description']}"
|
| 80 |
+
]),
|
| 81 |
+
metadata={"name": veg["name"]}
|
| 82 |
+
)
|
| 83 |
+
for veg in vegetable_data
|
| 84 |
+
]
|
| 85 |
+
|
| 86 |
+
# Return the tool
|
| 87 |
+
return VegetableInfoRetrieverTool(docs)
|