Instructions to use mtzig/qwen3-8b-tfdark-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mtzig/qwen3-8b-tfdark-lora with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-8B")
model = PeftModel.from_pretrained(base_model, "mtzig/qwen3-8b-tfdark-lora")

Transformers

How to use mtzig/qwen3-8b-tfdark-lora with Transformers:

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

pipe = pipeline("text-generation", model="mtzig/qwen3-8b-tfdark-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mtzig/qwen3-8b-tfdark-lora", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use mtzig/qwen3-8b-tfdark-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mtzig/qwen3-8b-tfdark-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mtzig/qwen3-8b-tfdark-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mtzig/qwen3-8b-tfdark-lora

SGLang

How to use mtzig/qwen3-8b-tfdark-lora 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 "mtzig/qwen3-8b-tfdark-lora" \
    --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": "mtzig/qwen3-8b-tfdark-lora",
		"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 "mtzig/qwen3-8b-tfdark-lora" \
        --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": "mtzig/qwen3-8b-tfdark-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use mtzig/qwen3-8b-tfdark-lora with Docker Model Runner:
```
docker model run hf.co/mtzig/qwen3-8b-tfdark-lora
```

qwen3-8b-tfdark-lora

This model is a fine-tuned version of Qwen/Qwen3-8B on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.4613

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0002
train_batch_size: 2
eval_batch_size: 2
seed: 4234
gradient_accumulation_steps: 8
total_train_batch_size: 16
optimizer: Use paged_adamw_32bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.03
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.7684	0.0390	10	0.6517
0.7136	0.0779	20	0.5949
0.6092	0.1169	30	0.5779
0.6011	0.1559	40	0.5638
0.5558	0.1948	50	0.6038
0.6974	0.2338	60	0.5189
0.548	0.2728	70	0.5190
0.5519	0.3117	80	0.4997
0.6448	0.3507	90	0.5379
0.6097	0.3897	100	0.4691
0.5982	0.4286	110	0.4946
0.5577	0.4676	120	0.5205
0.4735	0.5066	130	0.5119
0.531	0.5455	140	0.5022
0.594	0.5845	150	0.4483
0.4909	0.6235	160	0.4713
0.4483	0.6624	170	0.4820
0.5622	0.7014	180	0.4733
0.5791	0.7404	190	0.4495
0.5893	0.7793	200	0.4526
0.4988	0.8183	210	0.4563
0.5362	0.8573	220	0.4597
0.4293	0.8962	230	0.4597
0.4972	0.9352	240	0.4583
0.4901	0.9742	250	0.4613

Framework versions

PEFT 0.17.1
Transformers 4.56.2
Pytorch 2.8.0+cu128
Datasets 4.1.1
Tokenizers 0.22.1

Downloads last month: -

Model tree for mtzig/qwen3-8b-tfdark-lora

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Adapter

(1375)

this model