rshaikh22/Qwen3_30B_Medical

This is a merged model combining Qwen/Qwen3-30B-A3B-Instruct-2507 with a LoRA adapter fine-tuned on Japanese medical case studies.

Model Details

• Base Model: Qwen/Qwen3-30B-A3B-Instruct-2507
• Training Data: Japanese medical case studies (~93,563 examples)
• Fine-tuning Method: LoRA (Low-Rank Adaptation) - Merged
• Model Type: Merged Causal LM (no adapter needed)

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("rshaikh22/Qwen3_30B_Medical", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("rshaikh22/Qwen3_30B_Medical", trust_remote_code=True)

# Use the model
prompt = "Your prompt here"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=200)
print(tokenizer.decode(outputs[0]))

Training Details

• Epochs: 2
• Learning Rate: 5e-4
• Batch Size: 24
• Training Examples: ~93,563