Instructions to use zai-org/GLM-5.2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zai-org/GLM-5.2 with Transformers:

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

pipe = pipeline("text-generation", model="zai-org/GLM-5.2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("zai-org/GLM-5.2")
model = AutoModelForCausalLM.from_pretrained("zai-org/GLM-5.2")
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-5.2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zai-org/GLM-5.2"
# 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-5.2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/zai-org/GLM-5.2

SGLang

How to use zai-org/GLM-5.2 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-5.2" \
    --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-5.2",
		"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-5.2" \
        --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-5.2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use zai-org/GLM-5.2 with Docker Model Runner:
```
docker model run hf.co/zai-org/GLM-5.2
```

Vllm 0.23.0 fp8_e4m3 doesn't work on H200s

#10

by tkg61 - opened 9 days ago

Discussion

tkg61

9 days ago

VLLM returns an error when using fp8_e4m3 for kv-cache type. Have to use "fp8" instead.

Command: vllm serve zai-org/GLM-5.2-FP8 --trust-remote-code --tensor-parallel-size 8 --tool-call-parser glm47 --enable
-auto-tool-choice --reasoning-parser glm45 --served-model-name zai-org/GLM-5.2-FP8 --kv-cache-dtype fp8_e4m3 --gpu-memory-utilization 0.95

Error:
ValueError: No valid attention backend found for cuda with AttentionSelectorConfig(head_size=576, dtype=torch.bfloat16, kv_cache_dtype=fp8_e4m3, block_size=None, use
_mla=True, has_sink=False, use_sparse=True, use_mm_prefix=False, use_per_head_quant_scales=False, attn_type=AttentionType.DECODER, use_non_causal=False, use_batch_invariant=False, use_kv_connector=False). Reasons: {FLASH_ATTN_MLA: [kv_ca
che_dtype not supported, sparse not supported], FLASHMLA: [sparse not supported], FLASHINFER_MLA: [sparse not supported, compute capability not supported], TRITON_MLA: [sparse not supported], FLASHMLA_SPARSE: [kv_cache_dtype not supporte
d]}.

Also, H200s don't support a 1M context window unless that has something to do with using fp8 kv-cache. Could someone update the docs for H200s specifically?

g-a-b-y

9 days ago

@tkg61 you need B200's for 1M context. It is stated in the recipe guide here: https://recipes.vllm.ai/zai-org/GLM-5.2

For the H200's, you need 8 of them and context length of 512k.

JaredforReal

9 days ago

@tkg61 in H200, we recommend to use --kv-cache-dtype fp8_ds_mla for fp8 kv cache. and try to reduce concurrency like --max-num-seqs 32
thanks for your understanding

ZHANGYUXUAN-zR changed discussion status to closed 1 day ago

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