deshanksuman/finetuned-gemma-3-4b-it-WSD_Think

Uploaded model

• Developed by: deshanksuman
• License: apache-2.0
• Finetuned from model : google/gemma-3-4b-it

Dataset

Fews Training data arranged in the format of Instruction, Input and output deshanksuman/Instruct_Finetune_with_Reasoning_WSD

Hyperparameter for Training

• per_device_train_batch_size=4,
• gradient_accumulation_steps=8,
• warmup_steps=50,
• num_train_epochs=1,
• learning_rate=2e-4,
• fp16=not torch.cuda.is_bf16_supported(),
• bf16=torch.cuda.is_bf16_supported(),
• logging_steps=10,
• optim="adamw_torch",
• weight_decay=0.01,
• lr_scheduler_type="linear",
• seed=3407

This is developed by Deshan Sumanathilaka https://sumanathilaka.github.io

Acknowledgement

We acknowledge the support of the Supercomputing Wales project, which is part-funded by the European Regional Development Fund (ERDF) via Welsh Government.