jiteshsureka/gemma-3-1b-ecomm

Gemma 3 1B E-commerce Product Similarity Model

This model is a fine-tuned version of unsloth/gemma-3-1b-pt-unsloth-bnb-4bit for e-commerce product similarity tasks.

Model Details

• Base Model: unsloth/gemma-3-1b-pt
• Training Method: LoRA (Low-Rank Adaptation) with Unsloth
• Training Dataset: jiteshsureka/retail-ecomm-products
• Training Steps: 60
• Training Date: 2025-08-07

Usage

This model is designed to compare product similarity and provide scores between 0.0 and 1.0.

from unsloth import FastLanguageModel
import torch

# Load model
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="jiteshsureka/gemma-3-1b-ecomm",
    max_seq_length=2048,
    dtype=None,
    load_in_4bit=True,
)

# Enable inference mode
FastLanguageModel.for_inference(model)

# Example usage
prompt = '''<bos><start_of_turn>user
Compare these two products and rate their similarity from 0.0 to 1.0:

Product 1: Your product name
Description: Your product description
Category: Your category

Product 2: Another product name
Description: Another product description
Category: Another category

How similar are these products?<end_of_turn>
<start_of_turn>model
'''

inputs = tokenizer([prompt], return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=64, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

Training Configuration

• LoRA Rank: 16
• LoRA Alpha: 32
• Learning Rate: 2e-4
• Batch Size: 1
• Gradient Accumulation: 4

License

This model is released under the Apache 2.0 license.