This model is a fine-tuned version of
Qwen/Qwen2.5-7B-Instruct on FOMC (Federal Open Market Committee) and Beige Book data.
1import torch
2from transformers import AutoModelForCausalLM, AutoTokenizer
3from peft import PeftModel
4
5# Load base model
6base_model = AutoModelForCausalLM.from_pretrained(
7 "Qwen/Qwen2.5-7B-Instruct",
8 torch_dtype=torch.bfloat16,
9 device_map="auto",
10 trust_remote_code=True
11)
12
13# Load tokenizer
14tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-7B-Instruct", trust_remote_code=True)
15
16# Load LoRA adapter
17model = PeftModel.from_pretrained(
18 base_model,
19 "jaeyoungk/qwen-sft",
20 torch_dtype=torch.bfloat16,
21 device_map="auto"
22)
23
24# Sample usage
25messages = [
26 {"role": "user", "content": "What are the key risks to the economic outlook according to FOMC?"}
27]
28
29# Apply chat template
30prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
31inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
32
33with torch.no_grad():
34 outputs = model.generate(**inputs, max_length=2048, temperature=0.7, do_sample=True)
35
36response = tokenizer.decode(outputs[0], skip_special_tokens=True)
37print(response)
