Hugging Face's logo

amd
/
GLM-5-NVFP4

Safetensors
glm_moe_dsa
8-bit precision
quark
Model card Files
xet
 Community
5
GLM-5-NVFP4
File size: 3,187 Bytes
d7adec7
939c8f0
 
 
 
 
d7adec7
1063bab
939c8f0
1063bab
939c8f0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
---
license: other
license_name: modified-mit
license_link: LICENSE
base_model:
- zai-org/GLM-5
---

# Model Overview

- **Model Architecture:** GLM-5
  - **Input:** Text
  - **Output:** Text
- **Supported Hardware Microarchitecture:** AMD MI300/MI350/MI355 (emulation)
- **ROCm:** 7.2.2
- **Operating System(s):** Linux
- **Inference Engine:** [vLLM](https://docs.vllm.ai/en/latest/)
- **Model Optimizer:** [AMD-Quark](https://quark.docs.amd.com/latest/index.html) (V0.12)
  - **Weight quantization:** MOE-only, NVFP4, Static 
  - **Activation quantization:** MOE-only, NVFP4, Dynamic 
- **Calibration Dataset:** [Pile](https://huggingface.co/datasets/mit-han-lab/pile-val-backup)

This model was built with GLM-5 model by applying [AMD-Quark](https://quark.docs.amd.com/latest/index.html) for NVFP4 quantization.
# Model Quantization

The model was quantized from [zai-org/GLM-5](https://huggingface.co/zai-org/GLM-5) using [AMD-Quark](https://quark.docs.amd.com/latest/index.html). The weights and activations are quantized to NVFP4.

**Quantization scripts:**
```
sudo sysctl -w vm.max_map_count=4194304
cd Quark/examples/torch/language_modeling/llm_ptq/
export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
export MODEL_DIR=/zai-org/GLM-5
export output_dir=/amd/GLM-5-NVFP4
exclude_layers="*self_attn* *mlp.gate *lm_head *mlp.gate_proj *mlp.up_proj *mlp.down_proj"
python3 quantize_quark.py --model_dir $MODEL_DIR \
                          --quant_scheme nvfp4 \
                          --num_calib_data 128 \
                          --exclude_layers $exclude_layers \
                          --model_export hf_format \
                          --output_dir $output_dir \
                          --multi_gpu balanced
```

# Deployment
### Use with vLLM

This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend.

## Evaluation
The model was evaluated on GSM8K benchmarks. 

### Accuracy

<table>
  <tr>
   <td><strong>Benchmark</strong>
   </td>
   <td><strong>GLM-5 </strong>
   </td>
   <td><strong>GLM-5-NVFP4(this model) </strong>
   </td>
   <td><strong>Recovery</strong>
   </td>
  </tr>
  <tr>
   <td>GSM8K (flexible-extract)
   </td>
   <td>95.45
   </td>
   <td>95.22
   </td>
   <td>99.75%
   </td>
  </tr>

  </tr>
</table>

### Reproduction

The GSM8K result was obtained using the `lm-evaluation-harness` framework, based on the Docker image `rocm/vllm-dev:nightly_main_20260603`.

Install the lm-eval `(Version: 0.4.12)` in container first.
```
pip install lm-eval
pip install lm-eval[api]
```
#### Launching Server and Evaluating model
```
export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
export VLLM_ROCM_USE_AITER=1
export VLLM_ROCM_USE_AITER_MLA=1
export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
export PYTORCH_ALLOC_CONF=expandable_segments:True
lm_eval \
  --model vllm \
  --model_args pretrained=/amd/GLM-5-NVFP4,tensor_parallel_size=8,max_model_len=4096,gpu_memory_utilization=0.90,enforce_eager=True,max_gen_toks=2048,kv_cache_dtype=bfloat16,trust_remote_code=True \
  --tasks gsm8k \
  --num_fewshot 5 \
  --batch_size auto

```
```


# License
Modifications Copyright(c) 2026 Advanced Micro Devices, Inc. All rights reserved.