Instructions to use Er1ckRg/cuphead-qwen3-1.7b-finetuned with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Er1ckRg/cuphead-qwen3-1.7b-finetuned with Transformers:

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

pipe = pipeline("text-generation", model="Er1ckRg/cuphead-qwen3-1.7b-finetuned")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Er1ckRg/cuphead-qwen3-1.7b-finetuned")
model = AutoModelForCausalLM.from_pretrained("Er1ckRg/cuphead-qwen3-1.7b-finetuned", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Er1ckRg/cuphead-qwen3-1.7b-finetuned with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Er1ckRg/cuphead-qwen3-1.7b-finetuned"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Er1ckRg/cuphead-qwen3-1.7b-finetuned",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Er1ckRg/cuphead-qwen3-1.7b-finetuned

SGLang

How to use Er1ckRg/cuphead-qwen3-1.7b-finetuned 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 "Er1ckRg/cuphead-qwen3-1.7b-finetuned" \
    --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": "Er1ckRg/cuphead-qwen3-1.7b-finetuned",
		"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 "Er1ckRg/cuphead-qwen3-1.7b-finetuned" \
        --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": "Er1ckRg/cuphead-qwen3-1.7b-finetuned",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Er1ckRg/cuphead-qwen3-1.7b-finetuned with Docker Model Runner:
```
docker model run hf.co/Er1ckRg/cuphead-qwen3-1.7b-finetuned
```

cuphead-qwen3-1.7b-finetuned

Versão fine-tuned do Qwen/Qwen3-1.7B especializada em responder perguntas sobre o jogo Cuphead e sua DLC The Delicious Last Course, em português.

O modelo foi treinado com SFT (Supervised Fine-Tuning) usando TRL sobre o dataset Er1ckRg/cuphead_basico, cobrindo chefes, mecânicas, personagens, armas, amuletos e conteúdo da DLC.

Uso

from transformers import pipeline

generator = pipeline(
    "text-generation",
    model="Er1ckRg/cuphead-qwen3-1.7b-finetuned",
    device="cuda",
)

mensagens = [
    {
        "role": "system",
        "content": (
            "Você é um especialista no jogo Cuphead e sua DLC The Delicious Last Course. "
            "Responda perguntas de forma precisa e factual sobre a campanha, chefes, "
            "mecânicas e conteúdo do jogo."
        ),
    },
    {"role": "user", "content": "Quem são os protagonistas de Cuphead?"},
]

# Qwen3 gera tokens de thinking por padrão — filtre o bloco <think> se não quiser
output = generator(mensagens, max_new_tokens=512, do_sample=False)[0]
resposta = output["generated_text"][-1]["content"]

# Remove o bloco de raciocínio interno caso apareça
if "</think>" in resposta:
    resposta = resposta.split("</think>")[-1].strip()

print(resposta)

Dataset

Treinado sobre Er1ckRg/cuphead_basico, um dataset de perguntas e respostas factuais em português cobrindo:

Chefes da campanha principal e da DLC
Personagens jogáveis (Cuphead, Mugman, Ms. Chalice)
Armas, amuletos e mecânicas de jogo
Modo cooperativo e sistema de ranking

Formato: chat com roles system / user / assistant.

Procedimento de treino

Método: SFT (Supervised Fine-Tuning)
Loss: NLL apenas nos tokens do assistente (assistant_only_loss=True)
Épocas: 3
Hardware: NVIDIA Tesla T4 (16 GB)
Precisão: FP16

Versões dos frameworks

TRL: 1.7.0
Transformers: 5.12.1
PyTorch: 2.10.0+cu128
Datasets: 4.8.5
Tokenizers: 0.22.2

Limitações

Especializado em Cuphead — perguntas fora desse domínio serão respondidas com base no conhecimento geral do Qwen3-1.7B, sem garantia de qualidade.
Dataset de tamanho reduzido — pode não cobrir situações muito específicas ou raras do jogo.
Respostas em português; desempenho em outros idiomas não foi avaliado.

Citações

@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}
}

Downloads last month: 9

Safetensors

Model size

2B params

Tensor type

BF16

Model tree for Er1ckRg/cuphead-qwen3-1.7b-finetuned

Base model

Qwen/Qwen3-1.7B-Base

Finetuned

Qwen/Qwen3-1.7B

Finetuned

(1359)

this model