LiquidCode-2.5-24B-A2B

This model is a fine-tuned version of LiquidAI/LFM2.5-24B-A2B-Base using the dataset nvidia/Nemotron-SFT-SWE-v3. It has been trained using TRL and Unsloth.

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="ItachiSan/LiquidCode-2.5-24B-A2B", device_map="auto")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])

Training procedure

1. Take the model
2. Wire the dataset
3. Train
4. Profit!

No additional process was done as of now.

This model was trained with SFT.

Framework versions

• TRL: 1.15.0
• Transformers: 5.17.0
• Pytorch: 2.11.0
• Datasets: 4.8.5
• Tokenizers: 0.23.3

Citations

TRL:

@software{vonwerra2020trl,
  title   = {{TRL: Transformers Reinforcement Learning}},
  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
  license = {Apache-2.0},
  url     = {https://github.com/huggingface/trl},
  year    = {2020}
}