JT-Math-8B-Base

We are excited to introduce JT-Math-8B-Base: an 8-billion-parameter foundation model engineered for mathematical reasoning and the cornerstone of the JT-Math family. JT-Math-8B-Base was pre-trained on top of JT-Coder-8B-Base using an additional 210 billion tokens of high-quality mathematical and general-domain data. With a native 32,768-token context window, it provides a robust, scalable, and reproducible foundation for downstream fine-tuning—enabling researchers and developers to advance the frontier of math-centric AI applications. Technical details, training recipes, and reproducibility notes are available in our technical report.

Model Downloads

We release the following models to support a wide range of applications.

Model Name	Context Length	Hugging Face Link	ModelScope Link	Notes
JT-Math-8B-Base	32K	Link	Link	The foundational base model. Ideal for custom fine-tuning.

---

Evaluation Results

Model	GSM8K	Math	CMath (zh)	Average
Qwen2.5-Base-32B	92.8	57.7	85.4	78.6
Llama-3.1-Base-405B	89.0	53.8	77.4	73.4
DeepSeek-Math-Base-7B	64.2	36.2	71.7	57.4
DeepSeek-Coder-V2-Lite-Base	68.3	38.1	77.8	61.4
Qwen2.5-Math-7B	91.6	55.4	85.0	77.3
JT-Math-8B-Base	87.5	60.1	90.2	79.2

How to Get Started

We provide a basic example of how to run inference with the JT-Math-8B-Base model.

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "JT-LM/JT-Math-8B-Base"

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto",
    trust_remote_code=True,
)

prompt = "Janet’s ducks lay 16 eggs per day. She eats three for breakfast every morning and bakes muffins for her friends every day with four. She sells the remainder at the farmers' market daily for $2 per fresh duck egg. How much in dollars does she make every day at the farmers' market?"
text = f"Question:\n{prompt}\nAnswer:\n"
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

gen_kwargs = {
    "do_sample": False,
    "max_new_tokens": 8192,
}
generated_ids = model.generate(
    **model_inputs,
    **gen_kwargs
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()

response = tokenizer.decode(output_ids, skip_special_tokens=True)
print("response:", response)

Citation

If you find our work useful, please consider citing our paper:

@article{jiutian-math2025,
  title={JIUTIAN MATH: A MULTI-STAGE FRAMEWORK FOR ADVANCED MATHEMATICAL REASONING IN LARGE LANGUAGE MODELS},
  author={Yifan Hao, Fangning Chao, Yaqian Hao, Zhaojun Cui, Huan Bai, Haiyu Zhang, Yankai Liu, Chao Deng, Junlan Feng},
  journal={arXiv:2507.19748},
  year={2025}
}