EasySell AI β€” SLM Tool Calling Model

A custom GPT model trained to output structured JSON tool calls for a WhatsApp e-commerce shopping assistant.

Architecture

  • Type: Custom GPT (Decoder-only Transformer)
  • Layers: 14 | Heads: 12 | Embedding: 768
  • Block size: 1024 tokens
  • Trained on: Kaggle P100 GPU, 3 epochs, loss ~0.016

What it does

Given a customer message, outputs a JSON tool call selecting one of 12 tools:

Input : "i want bag pack"
Output: {"tool": "add_to_cart_tool", "parameters": {"product": "bag pack", "quantity": 1}}

Input : "checkout"
Output: {"tool": "get_checkout_link_tool", "parameters": {}}

Input : "where is my order"
Output: {"tool": "get_order_status_tool", "parameters": {}}

Supported Tools (12)

Tool Trigger
search_catalog_by_text_tool Product search, link requests
add_to_cart_tool Buy, order, add any product
get_checkout_link_tool Checkout, pay now
get_instant_checkout_link_tool Buy now, instant buy
list_all_products_tool Show products, catalog
update_order_address_tool Change delivery address
view_cart_tool My cart, show cart
update_cart_item_tool Change quantity in cart
remove_from_cart_tool Remove item from cart
clear_cart_tool Empty entire cart
clear_out_of_stock_items_tool Remove unavailable items
get_order_status_tool Order tracking, delivery status

Languages

Supports English and Hindi/Hinglish:

  • "mujhe shawl chahiye" β†’ add_to_cart_tool
  • "cart dikhao" β†’ view_cart_tool
  • "mobile ki link bhejo" β†’ search_catalog_by_text_tool

Author

Built by Tanishq Vemaraju (https://huggingface.co/TanishqV03)

Downloads last month
170
Safetensors
Model size
0.1B params
Tensor type
F32
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support