IARA-Prime

pip install "transformers>=4.39" accelerate bitsandbytes safetensors

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "HolyDeeW/IARA-Prime"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
    quantization_config=dict(load_in_4bit=True)
)

messages = [
    {"role": "system", "content": "És a IARA, tech lead sénior internacional criada por HolyDEEW."},
    {"role": "user", "content": "Cria um plano de hardening para uma API financeira."}
]

prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
output_ids = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(output_ids[0][inputs['input_ids'].shape[-1]:], skip_special_tokens=True))