Instructions to use zai-org/GLM-4-32B-0414 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zai-org/GLM-4-32B-0414 with Transformers:

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

pipe = pipeline("text-generation", model="zai-org/GLM-4-32B-0414")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("zai-org/GLM-4-32B-0414")
model = AutoModelForCausalLM.from_pretrained("zai-org/GLM-4-32B-0414", 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]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use zai-org/GLM-4-32B-0414 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zai-org/GLM-4-32B-0414"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zai-org/GLM-4-32B-0414",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/zai-org/GLM-4-32B-0414

SGLang

How to use zai-org/GLM-4-32B-0414 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 "zai-org/GLM-4-32B-0414" \
    --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": "zai-org/GLM-4-32B-0414",
		"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 "zai-org/GLM-4-32B-0414" \
        --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": "zai-org/GLM-4-32B-0414",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use zai-org/GLM-4-32B-0414 with Docker Model Runner:
```
docker model run hf.co/zai-org/GLM-4-32B-0414
```

Awesome model! Can we get a version with a larger context window?

#15

by seall0 - opened Apr 28, 2025

Discussion

seall0

Apr 28, 2025

Great model! I use it to make detailed minutes based on transcripts of really long meetings in portuguese and this absolutely blows mixtral and everything else out of the water. It leaves absolutely no detail behind. It gets a bit too "creative" sometimes but I reduced the temperature and it's better now. Awesome model!

The context window could be higher for my use case though, 64k would be perfect. Can we get that?

mratsim

May 2, 2025

Note sure why it's missing from the model card but the README on Github does have the info: https://github.com/THUDM/GLM-4/blob/main/README.md#handling-long-context-yarn

If the total input + output token count might exceed the model's native context length (mostly 32k for the GLM-4-0414 series), it is recommended to enable YaRN to achieve better long-context modeling capabilities. For supported frameworks, you can modify the corresponding config.json. Specifically, for GLM-Z1 series models, consider enabling YaRN (Rope Scaling) when the input length exceeds 8,192 tokens.

"rope_scaling": {
    "factor": 4.0,
    "original_max_position_embeddings": 32768,
    "type": "yarn"
}

For most user requests, if the input + output token count does not exceed the native context length, no modifications are needed.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment