Instructions to use winglian/qwen3-4b-math-kd with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use winglian/qwen3-4b-math-kd with Transformers:

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

pipe = pipeline("text-generation", model="winglian/qwen3-4b-math-kd")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("winglian/qwen3-4b-math-kd")
model = AutoModelForCausalLM.from_pretrained("winglian/qwen3-4b-math-kd")
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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use winglian/qwen3-4b-math-kd with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "winglian/qwen3-4b-math-kd"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "winglian/qwen3-4b-math-kd",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/winglian/qwen3-4b-math-kd

SGLang

How to use winglian/qwen3-4b-math-kd 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 "winglian/qwen3-4b-math-kd" \
    --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": "winglian/qwen3-4b-math-kd",
		"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 "winglian/qwen3-4b-math-kd" \
        --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": "winglian/qwen3-4b-math-kd",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use winglian/qwen3-4b-math-kd with Docker Model Runner:
```
docker model run hf.co/winglian/qwen3-4b-math-kd
```

See axolotl config

axolotl version: 0.10.0.dev0

base_model: Qwen/Qwen3-4B-Base

plugins:
  - axolotl.integrations.kd.KDPlugin
  - axolotl.integrations.liger.LigerPlugin

liger_rms_norm: true
liger_glu_activation: true

  # torch_compile: true

strict: false

kd_trainer: true
kd_ce_alpha: 0.05
kd_alpha: 0.95
kd_temperature: 2.0
kd_online_server: vllm
kd_online_server_base_url: http://localhost:8888/
kd_online_topk: 40

dataloader_prefetch_factor: 8
dataloader_num_workers: 2
dataloader_pin_memory: true

gc_steps: -1  # gc at the end of each epoch

chat_template: qwen3
datasets:
  - path: winglian/OpenThoughts-114k-math-correct
    type: chat_template
    split: train
    split_thinking: true
    eot_tokens:
      - "<|im_end|>"

dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./outputs/out-kd-4b

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

wandb_project: kd-4b-math
wandb_entity: axolotl-ai
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 2
micro_batch_size: 4
num_epochs: 2
optimizer: adamw_torch_fused
adam_beta2: 0.95
lr_scheduler: rex
learning_rate: 3e-5
max_grad_norm: 0.1
save_safetensors: true

train_on_inputs: false
group_by_length: false
bf16: true
tf32: true

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
logging_steps: 1
flash_attention: true

warmup_steps: 100
evals_per_epoch: 4
saves_per_epoch: 1
debug:
weight_decay: 0.0
special_tokens:
  eos_token: <|im_end|>
deepspeed: deepspeed_configs/zero2_torch_compile.json

outputs/out-kd-4b

This model is a fine-tuned version of Qwen/Qwen3-4B-Base on the winglian/OpenThoughts-114k-math-correct dataset.

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: 3e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 2
total_train_batch_size: 32
total_eval_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 2.0

Training results

Framework versions

Transformers 4.51.3
Pytorch 2.7.0+cu128
Datasets 3.5.1
Tokenizers 0.21.1

Downloads last month: 2

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for winglian/qwen3-4b-math-kd

Base model

Qwen/Qwen3-4B-Base

Finetuned

(286)

this model

winglian
/

qwen3-4b-math-kd

outputs/out-kd-4b

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Model tree for winglian/qwen3-4b-math-kd

Dataset used to train winglian/qwen3-4b-math-kd