atom-27b

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "vanta-research/atom-27b"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="bfloat16",
    device_map="auto"
)

messages = [
    {"role": "user", "content": "What kind of thinking would you like to do together?"}
]

inputs = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt"
).to(model.device)

outputs = model.generate(
    inputs,
    max_new_tokens=512,
    do_sample=True,
    top_k=64,
    top_p=0.95
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))