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="jnmrr/img_test_smol", 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: 0.19.1
Transformers: 4.53.2
Pytorch: 2.6.0+cu124
Datasets: 4.0.0
Tokenizers: 0.21.2

Citations

Cite TRL as:

bibtex

@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}}
}

Training Metrics

Metric	Value
num_tokens	2713392.0000
train_samples_per_second	1.0380
total_flos	4218509368009728.0000
eval_mean_token_accuracy	0.9305
train_steps_per_second	0.0330
train_runtime	1650.6545
eval_loss	0.2482
train_loss	0.2751
eval_steps_per_second	0.4510
eval_num_tokens	2518937.0000
mean_token_accuracy	0.9304
learning_rate	0.0001
grad_norm	0.1802
epoch	3.0000
eval_samples_per_second	1.7870
eval_runtime	175.1086
loss	0.2596