Fine-Tuning Language Models from Human Preferences
Paper • 1909.08593 • Published • 4
How to use trl-lib/ppo_tldr with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="trl-lib/ppo_tldr") # Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("trl-lib/ppo_tldr")
model = AutoModelForCausalLM.from_pretrained("trl-lib/ppo_tldr")How to use trl-lib/ppo_tldr with vLLM:
# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "trl-lib/ppo_tldr"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "trl-lib/ppo_tldr",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker model run hf.co/trl-lib/ppo_tldr
How to use trl-lib/ppo_tldr with SGLang:
# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
--model-path "trl-lib/ppo_tldr" \
--host 0.0.0.0 \
--port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "trl-lib/ppo_tldr",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker run --gpus all \
--shm-size 32g \
-p 30000:30000 \
-v ~/.cache/huggingface:/root/.cache/huggingface \
--env "HF_TOKEN=<secret>" \
--ipc=host \
lmsysorg/sglang:latest \
python3 -m sglang.launch_server \
--model-path "trl-lib/ppo_tldr" \
--host 0.0.0.0 \
--port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "trl-lib/ppo_tldr",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'How to use trl-lib/ppo_tldr with Docker Model Runner:
docker model run hf.co/trl-lib/ppo_tldr
This model is a fine-tuned version of None. It has been trained using TRL.
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="sergiopaniego/ppo_tldr", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
This model was trained with PPO, a method introduced in Fine-Tuning Language Models from Human Preferences.
Cite PPO as:
@article{mziegler2019fine-tuning,
title = {{Fine-Tuning Language Models from Human Preferences}},
author = {Daniel M. Ziegler and Nisan Stiennon and Jeffrey Wu and Tom B. Brown and Alec Radford and Dario Amodei and Paul F. Christiano and Geoffrey Irving},
year = 2019,
eprint = {arXiv:1909.08593}
}
Cite TRL as:
@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}
}