Instructions to use songff/SinglePO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use songff/SinglePO with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="songff/SinglePO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("songff/SinglePO")
model = AutoModelForCausalLM.from_pretrained("songff/SinglePO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use songff/SinglePO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "songff/SinglePO"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "songff/SinglePO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/songff/SinglePO

SGLang

How to use songff/SinglePO with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "songff/SinglePO" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "songff/SinglePO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

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 "songff/SinglePO" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "songff/SinglePO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use songff/SinglePO with Docker Model Runner:
```
docker model run hf.co/songff/SinglePO
```

songff commited on Jul 26, 2025

Commit

aa73958

verified ·

1 Parent(s): f58738a

Create tokenization.py

Browse files

Files changed (1) hide show

tokenization.py +53 -0

tokenization.py ADDED Viewed

	@@ -0,0 +1,53 @@

+from transformers import PreTrainedTokenizerFast
+class SinglePOTokenizer(PreTrainedTokenizerFast):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+    def get_context(
+        self,
+        raw_instruction,
+        rule_description,
+    ):
+        prompt = "You are an expert prompt engineer." + " "
+        prompt += "Please help me optimize this prompt to get better response:\n\n[The Start of Raw Prompt]\n{}\n[The End of Raw Prompt]".format(raw_instruction)
+        prompt += "\n\nYou should optimize this prompt by {}".format(rule_description)
+        context = self.apply_chat_template(
+            [
+                {
+                    "role": "user",
+                    "content": prompt,
+                }
+            ],
+            add_generation_prompt=True,
+            tokenize=False,
+        ) + "The Optimized Prompt:\n\n[The Start of Optimized Prompt"
+        return context
+    def parse_output(
+        self,
+        output_text,
+        raw_instruction = "", # recommend to provide, so when some error happened, we can still use the raw instruction
+    ):
+        better_instruction = "The Optimized Prompt:\n\n[The Start of Optimized Prompt" + output_text
+        if "[The Start of Optimized Prompt]" in better_instruction:
+            better_instruction = better_instruction[better_instruction.index("[The Start of Optimized Prompt]") + len("[The Start of Optimized Prompt]"):]
+            if better_instruction.startswith("\n"):
+                better_instruction = better_instruction[1:]
+        if "[The End of Optimized Prompt]" in better_instruction:
+            better_instruction = better_instruction[:better_instruction.index("[The End of Optimized Prompt]")]
+            if better_instruction.endswith("\n"):
+                better_instruction = better_instruction[:-1]
+        if "The Optimized Prompt:" in better_instruction: # almost error happened
+            better_instruction = better_instruction[:better_instruction.index("The Optimized Prompt:")]
+        if better_instruction.strip() == "": # some error may happen in optimization, so use the raw instruction
+            better_instruction = raw_instruction
+        if "The Optimized" in better_instruction: # still some error happened
+            better_instruction = raw_instruction
+        return better_instruction