You can deploy this model and start earning money today!

SuperHotDogCat/
jmmlu-gemma-3-12b-thinking

Name: jmmlu-gemma-3-12b-thinking
Availability: InStock

safetensors

gemma3

Views

No views yet

JMMLU GPRO trained gemma3 12b model

How to use

python

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "SuperHotDogCat/jmmlu-gemma-3-12b-thinking" 
tokenizer = AutoTokenizer.from_pretrained(model_name)
tokenizer.pad_token_id = tokenizer.eos_token_id
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
model.eval()

# gemma3p rompt
prompt_template = '''<start_of_turn>user
You are a helpful and intelligent AI assistant.
Please perform reasoning to answer the user's question.
First, think through the reasoning process step by step, then provide both the reasoning and the answer to the user.
Enclose the reasoning process and the answer in <think></think> and <answer></answer> tags respectively.
Please provide your response in the following format:
<think> Describe your reasoning process here </think>
<end_of_turn>
<start_of_turn>user
{user_input}
<end_of_turn>
<start_of_turn>model
'''

# 推論
with torch.no_grad():
    while True:
        input_text = input("input: ")
        input_text = prompt_template.format(user_input=input_text)
        inputs = tokenizer(input_text, return_tensors="pt")  
        outputs = model.generate(
            input_ids=inputs["input_ids"],
            attention_mask=inputs["attention_mask"],
            max_new_tokens=1024,
            top_k=64,
            top_p=0.95,
        )
        # decode
        result_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
        print(result_text)