qwen2.5-1.5b-dolly-lora

qwen2.5-1.5b-dolly-lora

LoRA adapter fine-tuning Qwen/Qwen2.5-1.5B for instruction following, trained on a curated, category-balanced subset of databricks-dolly-15k (4,000 examples, 500 per task category).

Trained locally on Apple Silicon (MPS) with peft + trl's SFTTrainer, using prompt/completion loss masking so the model only learns to generate responses, not to reproduce instructions.

See the curated training dataset and [training code](the training repo) for details.

Usage

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-1.5B")
model = PeftModel.from_pretrained(base_model, "sukrit-arora/qwen2.5-1.5b-dolly-lora")
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-1.5B")

prompt = """Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
Explain what a LoRA adapter is in one paragraph.

### Response:
"""
inputs = tokenizer(prompt, return_tensors="pt")
output = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(output[0], skip_special_tokens=True))