Model Card for granite_c4

SYS="GAME_MODE=CONNECT_FOUR "

Board should be passed in this format as USER

state:O

  A B C D E F G
6 . . . . . . .
5 . . . X . . .
4 . . O O . . .
3 . . X X O . .
2 O . O O X . X
1 X . O X X . O

USER=""state:O\n\n A B C D E F G\n6 . . . . . . .\n5 . . . . . . .\n4 . . . . . . .\n3 . . . O . . .\n2 . . . X . . .\n1 O . O X O X X\n""

This model is a fine-tuned version of ibm-granite/granite-3.1-2b-instruct. It has been trained using TRL.

Quick start

from transformers import pipeline

question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="Parsenal/granite_c4", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])

Training procedure

This model was trained with SFT.

Framework versions

• TRL: 0.17.0
• Transformers: 4.48.3
• Pytorch: 2.5.1
• Datasets: 3.3.2
• Tokenizers: 0.21.1

Citations

Cite TRL as:

@misc{vonwerra2022trl,
    title        = {{TRL: Transformer Reinforcement Learning}},
    author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
    year         = 2020,
    journal      = {GitHub repository},
    publisher    = {GitHub},
    howpublished = {\url{https://github.com/huggingface/trl}}
}