Instructions to use MiniLLM/VanillaKD-Pretrain-Qwen-1.2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MiniLLM/VanillaKD-Pretrain-Qwen-1.2B with Transformers:

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

pipe = pipeline("text-generation", model="MiniLLM/VanillaKD-Pretrain-Qwen-1.2B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MiniLLM/VanillaKD-Pretrain-Qwen-1.2B")
model = AutoModelForCausalLM.from_pretrained("MiniLLM/VanillaKD-Pretrain-Qwen-1.2B", 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 MiniLLM/VanillaKD-Pretrain-Qwen-1.2B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MiniLLM/VanillaKD-Pretrain-Qwen-1.2B

SGLang

How to use MiniLLM/VanillaKD-Pretrain-Qwen-1.2B 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 "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B" \
    --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": "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B",
		"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 "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B" \
        --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": "MiniLLM/VanillaKD-Pretrain-Qwen-1.2B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MiniLLM/VanillaKD-Pretrain-Qwen-1.2B with Docker Model Runner:
```
docker model run hf.co/MiniLLM/VanillaKD-Pretrain-Qwen-1.2B
```

# VanillaKD-Pretrain-Qwen-1.2B

[paper](https://arxiv.org/abs/2410.17215) | [code](https://github.com/thu-coai/MiniPLM)

**VanillaKD-Pretrain-Qwen-1.2B** is a 1.2B model with Qwen achitecture pre-trained with vanilla token-level knowledge distillation on [the Pile](https://huggingface.co/datasets/monology/pile-uncopyrighted) for 50B tokens. The teacher model is [Qwen1.5-1.8B](https://huggingface.co/Qwen/Qwen1.5-1.8B).

We also open-source the tokenized [pre-training corpus](https://huggingface.co/datasets/MiniLLM/pile-tokenized) for reproducibility.

**It is used as the baseline for [MiniLLM-Qwen-1.2B](https://huggingface.co/MiniLLM/MiniPLM-Qwen-1.2B)**

## Evaluation

MiniPLM models achieves better performance given the same computation and scales well across model sizes:

<p align='left'>
    <img src="https://cdn-uploads.huggingface.co/production/uploads/624ac662102fcdff87be51b9/EOYzajQcwQFT5PobqL3j0.png" width="1000">
</p>

## Other Baselines
+ [Conventional Pre-Training](https://huggingface.co/MiniLLM/Pretrain-Qwen-1.2B)

## Citation

```bibtext
@article{miniplm,
    title={MiniPLM: Knowledge Distillation for Pre-Training Language Models}, 
    author={Yuxian Gu and Hao Zhou and Fandong Meng and Jie Zhou and Minlie Huang},
    journal={arXiv preprint arXiv:2410.17215},
    year={2024}
}

Downloads last month: 19

Safetensors

Model size

1B params

Tensor type

F32

Datasets used to train MiniLLM/VanillaKD-Pretrain-Qwen-1.2B

Paper for MiniLLM/VanillaKD-Pretrain-Qwen-1.2B

MiniPLM: Knowledge Distillation for Pre-Training Language Models

Paper • 2410.17215 • Published Oct 22, 2024 • 16