Instructions to use Devikshah/Qwen3-4b_Fashion with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Devikshah/Qwen3-4b_Fashion with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Base")
model = PeftModel.from_pretrained(base_model, "Devikshah/Qwen3-4b_Fashion")

Transformers

How to use Devikshah/Qwen3-4b_Fashion with Transformers:

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

pipe = pipeline("text-generation", model="Devikshah/Qwen3-4b_Fashion")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Devikshah/Qwen3-4b_Fashion", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Devikshah/Qwen3-4b_Fashion with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Devikshah/Qwen3-4b_Fashion

SGLang

How to use Devikshah/Qwen3-4b_Fashion 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 "Devikshah/Qwen3-4b_Fashion" \
    --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": "Devikshah/Qwen3-4b_Fashion",
		"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 "Devikshah/Qwen3-4b_Fashion" \
        --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": "Devikshah/Qwen3-4b_Fashion",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Devikshah/Qwen3-4b_Fashion with Docker Model Runner:
```
docker model run hf.co/Devikshah/Qwen3-4b_Fashion
```

A newer version of this model is available: Qwen/Qwen3-4B-Base

Qwen/Qwen3-4B-Base Fashion Exclusion Inference Model

Model Overview

This model is a fine-tuned version of Qwen/Qwen3-4B-Base designed to identify incompatible fashion styles based on a user's profile. It uses QLoRA 4-bit fine-tuning for efficient training and memory usage. The model predicts negative style exclusions based on user preferences, helping recommendation systems avoid incompatible fashion styles.

Model Details

Model Description

This model is fine-tuned on a dataset that includes user preferences and lifestyle information, inferring 2–3 fashion styles that do not fit the user's profile. It assigns a Confidence Score (High, Medium, Low) to each exclusion and provides reasoning for each style exclusion.

Fine-tuned from model: Qwen/Qwen3-4B-Base

Model Sources

Repository: https://huggingface.co/Qwen/Qwen3-4B-Base

Uses

Direct Use

This model is intended to be used for filtering fashion styles based on user profiles, helping fashion recommendation systems identify styles that are incompatible with the user.

Downstream Use

This model can be integrated into larger fashion recommendation systems or style analysis tools, where it can continuously learn and adapt to user preferences and trends.

Out-of-Scope Use

This model should not be used for generating new fashion styles or recommending outfits. It is strictly for identifying negative style exclusions based on user profiles.

Bias, Risks, and Limitations

Bias

The model may exhibit bias based on the training data it was fine-tuned on, particularly if the dataset does not account for a diverse range of body types, cultural backgrounds, and fashion preferences. Special care should be taken to monitor for fairness and inclusivity in the results.

Risks

This model is designed to exclude fashion styles based on user profiles. However, exclusion recommendations may not always align with the user’s actual preferences or societal fashion trends, especially for users whose profiles are underrepresented.

Recommendations

Users should be made aware of the limitations of this model. It is essential to balance user exclusions with positive recommendations to maintain an inclusive and diverse experience.

How to Get Started with the Model

Use the code below to load the model and generate fashion style exclusions.

from transformers import AutoTokenizer, AutoModelForCausalLM

# Load the model and tokenizer
model_name = "Devikshah/qwen3-4b-qlora-output"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Example input profile
user_profile = "Gender: Male, Prefers structured outfits, Corporate executive, Dislikes bold graphics"

# Tokenize the input
inputs = tokenizer(user_profile, return_tensors="pt")

# Generate output
output = model.generate(inputs["input_ids"])

# Decode and print result
print(tokenizer.decode(output[0], skip_special_tokens=True))

- PEFT 0.18.1

Downloads last month: 2

Model tree for Devikshah/Qwen3-4b_Fashion

Base model

Qwen/Qwen3-4B-Base

Adapter

(87)

this model