import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
MODEL_PATH = "eridai/erida-hotei-50125"
tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
MODEL_PATH,
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
trust_remote_code=True
)
device = "cuda" if torch.cuda.is_available() else "cpu"
model = model.to(device)
prompt = "Почему трава зеленая?"
messages = [
{"role": "system", "content": "Ты — доброжелательный ассистент, отвечающий на русском языке."},
{"role": "user", "content": prompt}
]
input_ids = tokenizer.apply_chat_template(
messages,
add_generation_prompt=True,
return_tensors="pt",
enable_thinking=True # режим рассуждения
).to(device)
with torch.no_grad():
outputs = model.generate(
input_ids,
max_new_tokens=1024,
temperature=0.6,
repetition_penalty=1.1,
do_sample=True,
pad_token_id=tokenizer.pad_token_id,
eos_token_id=tokenizer.eos_token_id
)
response_ids = outputs[0][len(input_ids[0]):]
generated_text = tokenizer.decode(response_ids, skip_special_tokens=True)
print(generated_text.strip())