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

Libraries

How to use MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math with Transformers:

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

pipe = pipeline("text-generation", model="MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math")
model = AutoModelForCausalLM.from_pretrained("MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math

SGLang

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

Docker Model Runner
How to use MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math with Docker Model Runner:
```
docker model run hf.co/MLLM-CL/Qwen3-4B-Instruct-2507-SFT-Math
```

Qwen3-4B-Instruct-2507-SFT-Math / trainer_state.json

Moenupa

Upload folder using huggingface_hub

09ba4f6 verified about 1 month ago

Raw

History Blame Contribute Delete

1.78 kB

	{
	"best_global_step": null,
	"best_metric": null,
	"best_model_checkpoint": null,
	"epoch": 1.0,
	"eval_steps": 500,
	"global_step": 50,
	"is_hyper_param_search": false,
	"is_local_process_zero": true,
	"is_world_process_zero": true,
	"log_history": [
	{
	"epoch": 0.2,
	"grad_norm": 6.699235885703409,
	"learning_rate": 9.806308479691595e-06,
	"loss": 3.3526,
	"step": 10
	},
	{
	"epoch": 0.4,
	"grad_norm": 7.265192108764275,
	"learning_rate": 7.795964517353734e-06,
	"loss": 0.5978,
	"step": 20
	},
	{
	"epoch": 0.6,
	"grad_norm": 5.588145236673648,
	"learning_rate": 4.477357683661734e-06,
	"loss": 0.6108,
	"step": 30
	},
	{
	"epoch": 0.8,
	"grad_norm": 7.117965141671009,
	"learning_rate": 1.4033009983067454e-06,
	"loss": 0.6467,
	"step": 40
	},
	{
	"epoch": 1.0,
	"grad_norm": 7.806578743138265,
	"learning_rate": 1.2179748700879013e-08,
	"loss": 0.6008,
	"step": 50
	},
	{
	"epoch": 1.0,
	"step": 50,
	"total_flos": 350412570624.0,
	"train_loss": 1.1617375755310058,
	"train_runtime": 75.1787,
	"train_samples_per_second": 21.283,
	"train_steps_per_second": 0.665
	}
	],
	"logging_steps": 10,
	"max_steps": 50,
	"num_input_tokens_seen": 0,
	"num_train_epochs": 1,
	"save_steps": 500,
	"stateful_callbacks": {
	"TrainerControl": {
	"args": {
	"should_epoch_stop": false,
	"should_evaluate": false,
	"should_log": false,
	"should_save": true,
	"should_training_stop": true
	},
	"attributes": {}
	}
	},
	"total_flos": 350412570624.0,
	"train_batch_size": 4,
	"trial_name": null,
	"trial_params": null
	}