cmd_product_matcher_steel

This model is a fine-tuned version of microsoft/deberta-v3-xsmall on product aliases, edits, BoL data history and additional data.

Evaluation Results

Overall Metrics

• Loss: 0.0838
• Accuracy: 0.9800
• Macro F1 Score: 0.9627
• Weighted F1 Score: 0.9800
• Macro Precision: 0.9605
• Macro Recall: 0.9649

Per-Class Metrics

Class	Precision	Recall	F1-score	Support
Irrelevant	0.9830	0.9715	0.9772	66024
Scrap	0.8921	0.8769	0.8844	2291
Steel Bars/Billets	0.9818	0.9896	0.9857	14903
Steel Beams	0.9502	0.9854	0.9675	2731
Steel Coils	0.9874	0.9858	0.9866	30880
Steel Pipes	0.9852	0.9937	0.9895	58042
Steel Plate	0.9462	0.9512	0.9487	7748
Steel Rods	0.9500	0.9641	0.9570	6760
Steel Slab	0.9688	0.9663	0.9676	772
Accuracy: 0.9800

Confusion Matrix

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("cmd_product_matcher_steel")
model = AutoModelForSequenceClassification.from_pretrained("cmd_product_matcher_steel")

# Example usage
text = "STEEL BARS ASTM 4959, 23,000 MT"
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
predictions = outputs.logits.argmax(-1)