mata5764/gte-Qwen2-1.5B-instruct-myfi-v2

Security Name Resolution Model

This is a fine-tuned version of Alibaba-NLP/gte-Qwen2-1.5B-instruct for security name resolution tasks.

Model Description

• Base Model: Alibaba-NLP/gte-Qwen2-1.5B-instruct
• Task: Security name matching and resolution
• Training Data: Financial security names and paraphrases
• License: Apache 2.0

Usage

from sentence_transformers import SentenceTransformer

# Load the model
model = SentenceTransformer('mata5764/gte-Qwen2-1.5B-instruct-myfi-v2')

# Encode security names
queries = ["Reliance", "HDFC Bank", "TCS"]
embeddings = model.encode(queries)

# Use for similarity search
similarity = model.similarity(embeddings, embeddings)

Training Details

• Fine-tuned for security name resolution
• Optimized for Indian financial securities
• Case-insensitive matching
• Custom evaluation metrics

Files Included

• Model weights (safetensors format)
• Complete tokenizer with custom Qwen tokenization (GTE-compatible)
• Configuration files
• Training metadata
• All necessary files for vLLM deployment

Deployment

This model includes all necessary files for deployment with:

• SentenceTransformers
• vLLM
• Transformers library
• Custom inference pipelines