datumo/CAC-CoT

Model Card for Model ID

Model Details

Model Description

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

• —Developed by: Sunguk Choi, Yonghoon Kwon, Heondeuk Lee
• —Shared by: SelectStar/Datumo
• —Model type: Decoder-only language model (Causal LM)
• —Language(s) (NLP): English
• —License: Apache License 2.0
• —Finetuned from model: 🔧 Qwen-2.5-7b-it

Model Sources

• —Repository: https://github.com/selectstar-ai/CAC-CoT
• —Paper: https://arxiv.org/abs/2508.18743

Direct Use

• —Solving reasoning problems requiring chain-of-thought (CoT).
• —Educational tutoring, math/logic assistants, explainable QA.
• —Applications requiring interpretable reasoning with low latency.

Downstream Use [optional]

• —Fine-tuning for specific reasoning benchmarks such as GSM8K, StrategyQA, or S1-Bench.
• —Integration into larger RAG or tutoring systems.

Out-of-Scope Use

• —Non-English tasks.
• —Open-ended creative generation (e.g., fiction, poetry).

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("datumo/CAC-CoT")  # 🔧 Replace with your model path
tokenizer = AutoTokenizer.from_pretrained("datumo/CAC-CoT")

prompt = "Problem: If you have 3 apples and get 2 more, how many do you have?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Citation

BibTeX:

@misc{choi2025caccotconnectorawarecompactchainofthought,
  title={CAC-CoT: Connector-Aware Compact Chain-of-Thought for Efficient Reasoning Data Synthesis Across Dual-System Cognitive Tasks},
  author={Sunguk Choi and Yonghoon Kwon and Heondeuk Lee},
  year={2025},
  eprint={2508.18743},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url={https://arxiv.org/abs/2508.18743},
}

More Information

• —System-1: Fast, intuitive reasoning
• —System-2: Slow, logical reasoning
• —Connector phrase: Fixed phrases guiding logical flow (e.g., “Because of this,” “Then,” etc.)
• —ART: Average Reasoning Trace length

Model Card Authors

Sunguk Choi, Yonghoon Kwon, Heondeuk Lee