Qwen 3.5 2B
A small-parameter open-source language model in the Qwen series with 2 billion parameters, intended for fine-tuning on domain-specific tasks.
Timeline (last 90 days)
In the e-commerce product classification task, the fine-tuned Qwen 3.5 2B model achieved an accuracy of 69.8%, surpassing the 68.2% of the un-fine-tuned GPT-6 Luna.
Serverless LoRA fine-tuning of Qwen 3.5 2B was performed on the Crusoe Intelligence Foundry platform, and this training cost only about 45 cents.
The experiment used the following hyperparameters: batch size 16, LoRA rank 16, LoRA alpha 32, dropout 0.05, learning rate 0.0002, 1 epoch, warm-up ratio 0.05, and weight decay 0.
All Facts (3)
In the e-commerce product classification task, the fine-tuned Qwen 3.5 2B model achieved an accuracy of 69.8%, surpassing the 68.2% of the un-fine-tuned GPT-6 Luna.
50%UnverifiedServerless LoRA fine-tuning of Qwen 3.5 2B was performed on the Crusoe Intelligence Foundry platform, and this training cost only about 45 cents.
50%UnverifiedThe experiment used the following hyperparameters: batch size 16, LoRA rank 16, LoRA alpha 32, dropout 0.05, learning rate 0.0002, 1 epoch, warm-up ratio 0.05, and weight decay 0.
50%