ecommerce_ops · ecommerce · workflow
Mercari fine-tunes a 2B-parameter LLM with QLoRA to extract listing attributes, outperforming GPT-3.5 Turbo at 14× lower cost
Extracting structured attributes from Mercari's diverse user-generated listing descriptions was difficult due to wide seller variability, category-specific attribute definitions, and continuously evolving content. Commercial LLM APIs were prohibitively expensive at production scale, and conventional ML models required continuous retraining as attribute definitions changed.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Listing description input
A listing description and a list of attribute keys to extract are provided as input.
Tools used
QLoRAgemma-2b-itGPT-3.5 turbollama.cppHuggingFaceW&BSFTTrainerGCP
Outcome
The QLoRA-fine-tuned gemma-2b-it model is approximately 95% smaller than the original base model, achieves a BLEU score slightly more than five percentage points higher than GPT-3.5 turbo, and is estimated to cost more than 14 times less to run in production.
Results
Volumeapproximately 95% smaller
Cost replacedmore than 14 times
Grounding & classification
Source type: technical build writeup
23 fields verified against source quotes.
data extractionknowledge basemetric backednamed customertools describedworkflow describedecommerceaccuracy improvementcost reductiontechnical build writeupdata entry opsecommerce opsdocument to record