Kaveri-Instruct-v1-lora (Qwen2.5-1.5B-Instruct LoRA)

Fine-tuned adapter for Qwen/Qwen2.5-1.5B-Instruct using the Kaveri Autonomous Wizard Pipeline.

• Developer / Organization: shiva123782 / RiyaVibe
• Base Model: Qwen/Qwen2.5-1.5B-Instruct
• Adapter Type: LoRA (PEFT)
• LoRA Rank (r): 8
• LoRA Alpha: 16
• Target Modules: ['v_proj', 'q_proj', 'gate_proj', 'k_proj', 'down_proj', 'o_proj', 'up_proj']
• Precision: float16 / 4-bit NF4 quantized training

Usage

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model_id = "Qwen/Qwen2.5-1.5B-Instruct"
adapter_id = "shiva123782/Kaveri-Instruct-v1-lora"

tokenizer = AutoTokenizer.from_pretrained(base_model_id)
base_model = AutoModelForCausalLM.from_pretrained(
    base_model_id,
    torch_dtype=torch.float16,
    device_map="auto"
)
model = PeftModel.from_pretrained(base_model, adapter_id)

messages = [
    {"role": "user", "content": "Who developed you?"}
]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=64)
print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))