Instructions to use jadechoi/wizl_base_sl-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jadechoi/wizl_base_sl-4b with Transformers:

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

pipe = pipeline("text-generation", model="jadechoi/wizl_base_sl-4b")
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("jadechoi/wizl_base_sl-4b")
model = AutoModelForCausalLM.from_pretrained("jadechoi/wizl_base_sl-4b", 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 jadechoi/wizl_base_sl-4b with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/jadechoi/wizl_base_sl-4b

SGLang

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

Docker Model Runner
How to use jadechoi/wizl_base_sl-4b with Docker Model Runner:
```
docker model run hf.co/jadechoi/wizl_base_sl-4b
```

See axolotl config

axolotl version: 0.12.2

base_model: Qwen/Qwen3-4B
load_in_8bit: false
load_in_4bit: false
strict: false

# Liger Kernel
plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_layer_norm: true
liger_fused_linear_cross_entropy: true

# Dataset
chat_template: qwen3
chat_template_kwargs:
  enable_thinking: false
datasets:
  - path: train_sl_v1.jsonl
    type: chat_template
    field_messages: messages
    message_property_mappings:
      role: role
      content: content
dataset_prepared_path: last_run_prepared
val_set_size: 0.03

# Output
output_dir: ./outputs/qwen3-4b-sl

# Full finetuning (no adapter)
adapter:
lora_model_dir:

# Sequence (SL 데이터 max ~493 chars / ~800 tokens → 1024 로 충분)
sequence_len: 1024
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

# Attention
flash_attention: true

# W&B
wandb_project: wizl-base-m
wandb_entity:
wandb_watch:
wandb_name: 4b-base-sl-v1
wandb_log_model:

# Hub
hub_model_id: jadechoi/wizl_base_sl-4b

# Training hyperparams
micro_batch_size: 16                 # SL 데이터 짧음(seq_len 1024) → QTA 대비 배치 2배
gradient_accumulation_steps: 2       # 2GPU × 16bs × 2accum = effective batch 64
num_epochs: 3
optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 2e-5
warmup_ratio: 0.05
weight_decay: 0.01

# Precision
bf16: true
fp16: false
tf32: true
gradient_checkpointing: false        # 4B + 짧은 seq → 비활성화로 속도 확보

# Logging & Saving
logging_steps: 1
evals_per_epoch: 1
saves_per_epoch: 1

# DeepSpeed ZeRO-2
deepspeed: ds_config.json

special_tokens:

wizl_base_sl-4b

This model is a fine-tuned version of Qwen/Qwen3-4B on the train_sl_v1.jsonl dataset. It achieves the following results on the evaluation set:

Loss: 0.0022
Memory/max Mem Active(gib): 129.1
Memory/max Mem Allocated(gib): 129.1
Memory/device Mem Reserved(gib): 136.63

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: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
distributed_type: multi-GPU
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 3
training_steps: 69

Training results

Training Loss	Epoch	Step	Validation Loss	Mem Active(gib)	Mem Allocated(gib)	Mem Reserved(gib)
No log	0	0	4.6729	46.36	46.36	46.47
0.0259	1.0	23	0.0301	129.1	129.1	136.63
0.0024	2.0	46	0.0037	129.1	129.1	136.63
0.0008	3.0	69	0.0022	129.1	129.1	136.63

Framework versions

Transformers 4.55.2
Pytorch 2.6.0+cu124
Datasets 4.0.0
Tokenizers 0.21.4

Downloads last month: 5

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for jadechoi/wizl_base_sl-4b

Base model

Qwen/Qwen3-4B-Base

Finetuned

Qwen/Qwen3-4B

Finetuned

(1223)

this model