xc-llm-ascend/docs/source/tutorials/models/GLM5.md

GLM-5

Introduction

GLM-5use a Mixture-of-Experts (MoE) architecture and targeting at complex systems engineering and long-horizon agentic tasks.

This document will show the main verification steps of the model, including supported features, feature configuration, environment preparation, single-node and multi-node deployment, accuracy and performance evaluation.

Supported Features

Refer to supported featuresto get the model's supported feature matrix.

Refer to feature guide to get the feature's configuration.

Environment Preparation

Model Weight

• GLM-5(BF16 version): Download model weight.
• GLM-5-w4a8(Quantized version without mtp): Download model weight.
• You can use msmodelslim to quantify the model naively.

It is recommended to download the model weight to the shared directory of multiple nodes, such as /root/.cache/

Installation

vLLM and vLLM-ascend only support GLM-5 on our main branches. you can use our official docker images and upgrade vllm and vllm-ascend for inference.

# Update --device according to your device (Atlas A3:/dev/davinci[0-15]).
# Update the vllm-ascend image according to your environment.
# Note you should download the weight to /root/.cache in advance.
# Update the vllm-ascend image, alm5-a3 can be replaced by: glm5;glm5-openeuler;glm5-a3-openeuler
export IMAGE=m.daocloud.io/quay.io/ascend/vllm-ascend:glm5-a3
export NAME=vllm-ascend

# Run the container using the defined variables
# Note: If you are running bridge network with docker, please expose available ports for multiple nodes communication in advance
docker run --rm \
--name $NAME \
--net=host \
--shm-size=1g \
--device /dev/davinci0 \
--device /dev/davinci1 \
--device /dev/davinci2 \
--device /dev/davinci3 \
--device /dev/davinci4 \
--device /dev/davinci5 \
--device /dev/davinci6 \
--device /dev/davinci7 \
--device /dev/davinci8 \
--device /dev/davinci9 \
--device /dev/davinci10 \
--device /dev/davinci11 \
--device /dev/davinci12 \
--device /dev/davinci13 \
--device /dev/davinci14 \
--device /dev/davinci15 \
--device /dev/davinci_manager \
--device /dev/devmm_svm \
--device /dev/hisi_hdc \
-v /usr/local/dcmi:/usr/local/dcmi \
-v /usr/local/Ascend/driver/tools/hccn_tool:/usr/local/Ascend/driver/tools/hccn_tool \
-v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
-v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ \
-v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \
-v /etc/ascend_install.info:/etc/ascend_install.info \
-v /root/.cache:/root/.cache \
-it $IMAGE bash

In addition, if you don't want to use the docker image as above, you can also build all from source:

• Install vllm-ascend from source, refer to installation.

To inference GLM-5, you should upgrade vllm、vllm-ascend、transformers to main branches:

# upgrade vllm
git clone https://github.com/vllm-project/vllm.git
cd vllm
git checkout 978a37c82387ce4a40aaadddcdbaf4a06fc4d590
VLLM_TARGET_DEVICE=empty pip install -v .

# upgrade vllm-ascend
git clone https://github.com/vllm-project/vllm-ascend.git
cd vllm-ascend
git checkout ff3a50d011dcbea08f87ebed69ff1bf156dbb01e
git submodule update --init --recursive
pip install -v .

# reinstall transformers
pip install git+https://github.com/huggingface/transformers.git

If you want to deploy multi-node environment, you need to set up environment on each node.

Deployment

Single-node Deployment

A2 series

Not test yet.

A3 series

• Quantized model glm-5-w4a8 can be deployed on 1 Atlas 800 A3 (64G × 16) .

Run the following script to execute online inference.

export HCCL_OP_EXPANSION_MODE="AIV"
export OMP_PROC_BIND=false
export OMP_NUM_THREADS=10
export VLLM_USE_V1=1
export HCCL_BUFFSIZE=200
export PYTORCH_NPU_ALLOC_CONF=expandable_segments:True
export VLLM_ASCEND_BALANCE_SCHEDULING=1

vllm serve /root/.cache/modelscope/hub/models/vllm-ascend/GLM5-w4a8 \
--host 0.0.0.0 \
--port 8077 \
--data-parallel-size 1 \
--tensor-parallel-size 16 \
--enable-expert-parallel \
--seed 1024 \
--served-model-name glm-5 \
--max-num-seqs 8 \
--max-model-len 66600 \
--max-num-batched-tokens 4096 \
--trust-remote-code \
--gpu-memory-utilization 0.95 \
--quantization ascend \
--enable-chunked-prefill \
--enable-prefix-caching \
--async-scheduling \
--additional-config '{"multistream_overlap_shared_expert":true}' \
--compilation-config '{"cudagraph_mode": "FULL_DECODE_ONLY"}' \
--speculative-config '{"num_speculative_tokens": 3, "method": "deepseek_mtp"}' 

Notice: The parameters are explained as follows:

• For single-node deployment, we recommend using dp1tp16 and turn off expert parallel in low-latency scenarios.
• --async-scheduling Asynchronous scheduling is a technique used to optimize inference efficiency. It allows non-blocking task scheduling to improve concurrency and throughput, especially when processing large-scale models.

Multi-node Deployment

A2 series

Not test yet.

A3 series

• glm-5-bf16: require at least 2 Atlas 800 A3 (64G × 16).

Run the following scripts on two nodes respectively.

node 0

# this obtained through ifconfig
# nic_name is the network interface name corresponding to local_ip of the current node
nic_name="xxx"
local_ip="xxx"

# The value of node0_ip must be consistent with the value of local_ip set in node0 (master node)
node0_ip="xxxx"

export HCCL_OP_EXPANSION_MODE="AIV"

export HCCL_IF_IP=$local_ip
export GLOO_SOCKET_IFNAME=$nic_name
export TP_SOCKET_IFNAME=$nic_name
export HCCL_SOCKET_IFNAME=$nic_name
export OMP_PROC_BIND=false
export OMP_NUM_THREADS=10
export VLLM_USE_V1=1
export HCCL_BUFFSIZE=200
export PYTORCH_NPU_ALLOC_CONF=expandable_segments:True

vllm serve /root/.cache/modelscope/hub/models/vllm-ascend/GLM5-bf16 \
--host 0.0.0.0 \
--port 8077 \
--data-parallel-size 2 \
--data-parallel-size-local 1 \
--data-parallel-address $node0_ip \
--data-parallel-rpc-port 12890 \
--tensor-parallel-size 16 \
--seed 1024 \
--served-model-name glm-5 \
--enable-expert-parallel \
--max-num-seqs 16 \
--max-model-len 8192 \
--max-num-batched-tokens 4096 \
--trust-remote-code \
--no-enable-prefix-caching \
--gpu-memory-utilization 0.95 \
--compilation-config '{"cudagraph_mode": "FULL_DECODE_ONLY"}' \
--speculative-config '{"num_speculative_tokens": 3, "method": "deepseek_mtp"}'

node 1

# this obtained through ifconfig
# nic_name is the network interface name corresponding to local_ip of the current node
nic_name="xxx"
local_ip="xxx"

# The value of node0_ip must be consistent with the value of local_ip set in node0 (master node)
node0_ip="xxxx"

export HCCL_OP_EXPANSION_MODE="AIV"

export HCCL_IF_IP=$local_ip
export GLOO_SOCKET_IFNAME=$nic_name
export TP_SOCKET_IFNAME=$nic_name
export HCCL_SOCKET_IFNAME=$nic_name
export OMP_PROC_BIND=false
export OMP_NUM_THREADS=10
export VLLM_USE_V1=1
export HCCL_BUFFSIZE=200
export PYTORCH_NPU_ALLOC_CONF=expandable_segments:True

vllm serve /root/.cache/modelscope/hub/models/vllm-ascend/GLM5-bf16 \
--host 0.0.0.0 \
--port 8077 \
--headless \
--data-parallel-size 2 \
--data-parallel-size-local 1 \
--data-parallel-start-rank 1 \
--data-parallel-address $node0_ip \
--data-parallel-rpc-port 12890 \
--tensor-parallel-size 16 \
--seed 1024 \
--served-model-name glm-5 \
--enable-expert-parallel \
--max-num-seqs 16 \
--max-model-len 8192 \
--max-num-batched-tokens 4096 \
--trust-remote-code \
--no-enable-prefix-caching \
--gpu-memory-utilization 0.95 \
--compilation-config '{"cudagraph_mode": "FULL_DECODE_ONLY"}' \
--speculative-config '{"num_speculative_tokens": 3, "method": "deepseek_mtp"}'

Prefill-Decode Disaggregation

Not test yet.

Accuracy Evaluation

Here are two accuracy evaluation methods.

Using AISBench

1. Refer to Using AISBench for details.

2. After execution, you can get the result.

Using Language Model Evaluation Harness

Not test yet.

Performance

Using AISBench

Refer to Using AISBench for performance evaluation for details.

Using vLLM Benchmark

Refer to vllm benchmark for more details.