from transformers import AutoModelForCausalLM, AutoTokenizer

model_path = "Ljy2004/mapfinben-qwen35-9b-merged-unified-v3"
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_path, trust_remote_code=True, torch_dtype="auto", device_map="auto"
)

messages = [{"role": "user", "content": "YOUR_PROMPT"}]
text = tokenizer.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True, enable_thinking=False
)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)

Important: Use Qwen3.5 chat template with enable_thinking=False to match training and official evaluation.

Citation

Base model: <a href="https://huggingface.co/Qwen/Qwen3.5-9B-Base" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">Qwen3.5-9B-Base
Method: LoRA SFT (rank=16, 1.5 epoch, lr=1e-4, cutoff_len=2048)
Training data: MapFinBen train split, unified 51,064 samples
Framework: LLaMA-Factory
Parameters: ~9.41B (merged full weights, ≤10B)

MapFinBen benchmark: https://github.com/HgITSE/MapFinBen