Instructions to use hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933 with Transformers:

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

pipe = pipeline("text-generation", model="hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933")
model = AutoModelForCausalLM.from_pretrained("hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933", device_map="auto")
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 Settings

vLLM

How to use hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

SGLang

How to use hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933 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 "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933" \
    --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": "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933",
		"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 "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933" \
        --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": "hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933 with Docker Model Runner:
```
docker model run hf.co/hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933
```

dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

This model is a fine-tuned version of Qwen/Qwen3-4B-Instruct-2507 using Direct Preference Optimization (DPO) via Unsloth.

Initialization path

Started from base model: Qwen/Qwen3-4B-Instruct-2507
Loaded SFT LoRA: hirosan6595/lora_structeval_t_qwen3_4b-21
Merged SFT LoRA into base first, then applied DPO LoRA training.

Training Configuration

Learning rate: 1e-07
Beta: 0.05
Epochs: 0.5
Max length: 2048
Gradient accumulation: 8

Notes

This repository contains merged_16bit weights uploaded by push_to_hub_merged.

Downloads last month: 6

Safetensors

Model size

4B params

Tensor type

F16

Model tree for hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(1957)

this model

hirosan6595
/

dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

Initialization path

Training Configuration

Notes

Model tree for hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933

Dataset used to train hirosan6595/dpo-qwen-cot-merged-lr1e-7_b0.05_ep0.5_ga8-20260216-202933