Instructions to use CEIA-RL/Qwen3-4B-Instruct-2507 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use CEIA-RL/Qwen3-4B-Instruct-2507 with Transformers:

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

pipe = pipeline("text-generation", model="CEIA-RL/Qwen3-4B-Instruct-2507")
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("CEIA-RL/Qwen3-4B-Instruct-2507")
model = AutoModelForCausalLM.from_pretrained("CEIA-RL/Qwen3-4B-Instruct-2507", 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 CEIA-RL/Qwen3-4B-Instruct-2507 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "CEIA-RL/Qwen3-4B-Instruct-2507"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "CEIA-RL/Qwen3-4B-Instruct-2507",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/CEIA-RL/Qwen3-4B-Instruct-2507

SGLang

How to use CEIA-RL/Qwen3-4B-Instruct-2507 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 "CEIA-RL/Qwen3-4B-Instruct-2507" \
    --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": "CEIA-RL/Qwen3-4B-Instruct-2507",
		"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 "CEIA-RL/Qwen3-4B-Instruct-2507" \
        --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": "CEIA-RL/Qwen3-4B-Instruct-2507",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use CEIA-RL/Qwen3-4B-Instruct-2507 with Docker Model Runner:
```
docker model run hf.co/CEIA-RL/Qwen3-4B-Instruct-2507
```

Model Card for Model ID


|    | model_name             |   task_coverage@1 |   task_coverage@2 |   task_coverage@4 |   task_coverage@8 |   relative_quality@1 |   relative_quality@2 |   relative_quality@4 |   relative_quality@8 |   hallucination@1 |   hallucination@2 |   hallucination@4 |   hallucination@8 |   final_score |
|---:|:-----------------------|------------------:|------------------:|------------------:|------------------:|---------------------:|---------------------:|---------------------:|---------------------:|------------------:|------------------:|------------------:|------------------:|--------------:|
|  9 | step_100               |          0.964928 |          0.978359 |          0.984942 |          0.98846  |             0.880293 |             0.903613 |             0.919117 |             0.929453 |          0.130947 |          0.102231 |         0.0858685 |         0.0716891 |       1.71427 |
|  7 | step_80                |          0.964611 |          0.975734 |          0.983325 |          0.986612 |             0.879933 |             0.900648 |             0.915331 |             0.927375 |          0.131163 |          0.102831 |         0.0859165 |         0.071737  |       1.71338 |
|  8 | step_90                |          0.96333  |          0.979069 |          0.984753 |          0.990053 |             0.872457 |             0.897927 |             0.91393  |             0.927558 |          0.140955 |          0.104607 |         0.0841651 |         0.0716891 |       1.69483 |
|  4 | step_50                |          0.963215 |          0.977606 |          0.983556 |          0.987793 |             0.870902 |             0.900749 |             0.918071 |             0.929707 |          0.143702 |          0.105998 |         0.0851248 |         0.0728887 |       1.69042 |
|  5 | step_60                |          0.963191 |          0.976468 |          0.98356  |          0.987529 |             0.876378 |             0.903704 |             0.920038 |             0.931228 |          0.151035 |          0.11166  |         0.087596  |         0.0720393 |       1.68853 |
|  2 | step_30                |          0.94999  |          0.965926 |          0.975245 |          0.980072 |             0.857476 |             0.880293 |             0.897927 |             0.911171 |          0.119615 |          0.102063 |         0.0871641 |         0.0769194 |       1.68785 |
|  6 | step_70                |          0.964007 |          0.976607 |          0.983301 |          0.987817 |             0.875456 |             0.901272 |             0.916843 |             0.928407 |          0.152841 |          0.119602 |         0.0933541 |         0.0795825 |       1.68662 |
|  3 | step_40                |          0.965302 |          0.980154 |          0.984755 |          0.988546 |             0.873369 |             0.900561 |             0.916867 |             0.928983 |          0.161827 |          0.119146 |         0.0934981 |         0.0774472 |       1.67684 |
|  1 | step_20                |          0.957207 |          0.971411 |          0.979069 |          0.982821 |             0.857438 |             0.880878 |             0.898201 |             0.908109 |          0.160553 |          0.127207 |         0.106502  |         0.09369   |       1.65409 |
| 11 | Qwen3-4B               |          0.955043 |          0.979347 |          0.990235 |          0.995178 |             0.859573 |             0.907198 |             0.93047  |             0.949136 |          0.193807 |          0.10763  |         0.0674664 |         0.0417946 |       1.62081 |
|  0 | step_10                |          0.949463 |          0.966785 |          0.975964 |          0.982908 |             0.843642 |             0.872313 |             0.891363 |             0.904774 |          0.189741 |          0.15096  |         0.125624  |         0.106958  |       1.60336 |
| 10 | Qwen3-4B-Instruct-2507 |          0.946771 |          0.960571 |          0.973853 |          0.981881 |             0.831468 |             0.864074 |             0.888018 |             0.904151 |          0.220855 |          0.171569 |         0.137706  |         0.114794  |       1.55738 |
| 12 | qwen3-4b-dw-lr         |          0.906166 |          0.96346  |          0.984189 |          0.99345  |             0.734621 |             0.833205 |             0.889002 |             0.925326 |          0.41     |          0.265427 |         0.171123  |         0.112644  |       1.23079 |

final_score = 'task_coverage@1' + 'relative_quality@1' -  'hallucination@1'

wandb: https://api.wandb.ai/links/fazzioni/1ytairgh

Comparação entre Checkpoints:

Comparação entre modelos:

Prompt utilizado no treinamento:

system_prompt = """Você é um auditor especializado em regulação do setor elétrico.
Sua tarefa é avaliar a fidelidade técnica de uma resposta baseada em um documento fonte e um gabarito (label).

Regras de Avaliação:
1. Verifique se a Resposta do LLM é suportada pelo documento fornecido em {source}.
2. Se a resposta contiver fundamentação técnica (Artigos, Resoluções) presente na fonte, atribua pontuação máxima, mesmo que o Label seja simplificado.
3. Se a resposta contradizer a fonte ou inventar dados não presentes, a nota deve ser 0.
4. Não penalize informações extras verídicas que complementem o Label.

Responda apenas com JSON:
{
  "raciocinio": "breve explicação da nota",
  "pontuacao": 0 a 10
}"""



prompt= """# <Pergunta>
    
    # documento que originou a pergunta e que pode conter informações relevantes para a resposta geradas do LLM.
    {source}
    
    
    # Pergunta
    {question}

    # <Label>
    {ground_truth}
    
    # <Resposta do LLM>
    {answer}"""

Model Details

Model Description

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

Developed by: [More Information Needed]
Funded by [optional]: [More Information Needed]
Shared by [optional]: [More Information Needed]
Model type: [More Information Needed]
Language(s) (NLP): [More Information Needed]
License: [More Information Needed]
Finetuned from model [optional]: [More Information Needed]

Model Sources [optional]

Repository: [More Information Needed]
Paper [optional]: [More Information Needed]
Demo [optional]: [More Information Needed]

Uses

Direct Use

[More Information Needed]

Downstream Use [optional]

[More Information Needed]

Out-of-Scope Use

[More Information Needed]

Bias, Risks, and Limitations

[More Information Needed]

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

[More Information Needed]

Training Procedure

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

Training regime: [More Information Needed]

Speeds, Sizes, Times [optional]

[More Information Needed]

Evaluation

Testing Data, Factors & Metrics

Testing Data

[More Information Needed]

Factors

[More Information Needed]

Metrics

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

[More Information Needed]

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Hardware Type: [More Information Needed]
Hours used: [More Information Needed]
Cloud Provider: [More Information Needed]
Compute Region: [More Information Needed]
Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]

Downloads last month: 8

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for CEIA-RL/Qwen3-4B-Instruct-2507

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(2724)

this model

Paper for CEIA-RL/Qwen3-4B-Instruct-2507

Quantifying the Carbon Emissions of Machine Learning

Paper • 1910.09700 • Published Oct 21, 2019 • 63