Instructions to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("unsloth/Qwen3.5-4B")
model = PeftModel.from_pretrained(base_model, "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari")

Transformers

How to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari with Transformers:

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

pipe = pipeline("text-generation", model="guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari

SGLang

How to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari 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 "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari" \
    --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": "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari",
		"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 "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari" \
        --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": "guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari with Docker Model Runner:
```
docker model run hf.co/guzus/screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari
```

screenwriter-qwen35-4b-dpo-smoke-sgs-ferrari / training_args.json

guzus

Upload smoke DPO LoRA adapter from SGS ferrari run

312ceab verified 3 months ago

Raw

History Blame

547 Bytes

	{
	"model": "/data/screenwriter/training/outputs/sgs-smoke/sft",
	"base_model": null,
	"seq_len": 2048,
	"load_in_4bit": false,
	"beta": 0.1,
	"data": "/data/screenwriter/training/outputs/_smoke_data/dpo_train.jsonl",
	"eval": "/data/screenwriter/training/outputs/_smoke_data/dpo_test.jsonl",
	"epochs": 1,
	"batch_size": 1,
	"grad_accum": 1,
	"lr": 2e-05,
	"warmup_steps": 0,
	"max_steps": 1,
	"output": "/data/screenwriter/training/outputs/sgs-smoke/dpo",
	"save_steps": 1,
	"logging_steps": 1,
	"seed": 3407,
	"wandb": false
	}