Update README.md
Browse files
README.md
CHANGED
|
@@ -7,20 +7,20 @@ tags:
|
|
| 7 |
- 量化修复
|
| 8 |
- vLLM
|
| 9 |
base_model:
|
| 10 |
-
-
|
| 11 |
base_model_relation: quantized
|
| 12 |
---
|
| 13 |
# GLM-4.5-AWQ
|
| 14 |
-
|
| 15 |
|
| 16 |
|
| 17 |
-
### 【
|
| 18 |
-
<i
|
| 19 |
```
|
| 20 |
CONTEXT_LENGTH=32768
|
| 21 |
|
| 22 |
vllm serve \
|
| 23 |
-
|
| 24 |
--served-model-name GLM-4.5-AWQ \
|
| 25 |
--enable-expert-parallel \
|
| 26 |
--swap-space 16 \
|
|
@@ -35,69 +35,69 @@ vllm serve \
|
|
| 35 |
--port 8000
|
| 36 |
```
|
| 37 |
|
| 38 |
-
###
|
| 39 |
|
| 40 |
```
|
| 41 |
vllm==0.10.0
|
| 42 |
```
|
| 43 |
|
| 44 |
-
###
|
| 45 |
```
|
| 46 |
2025-07-30
|
| 47 |
-
1.
|
| 48 |
```
|
| 49 |
|
| 50 |
-
###
|
| 51 |
|
| 52 |
-
|
|
| 53 |
|---------|--------------|
|
| 54 |
| `181GB` | `2025-07-30` |
|
| 55 |
|
| 56 |
|
| 57 |
|
| 58 |
-
###
|
| 59 |
|
| 60 |
```python
|
| 61 |
-
from
|
| 62 |
-
snapshot_download('
|
| 63 |
```
|
| 64 |
|
| 65 |
|
| 66 |
-
###
|
| 67 |
# GLM-4.5
|
| 68 |
|
| 69 |
<div align="center">
|
| 70 |
<img src=https://raw.githubusercontent.com/zai-org/GLM-4.5/refs/heads/main/resources/logo.svg width="15%"/>
|
| 71 |
</div>
|
| 72 |
<p align="center">
|
| 73 |
-
👋
|
| 74 |
<br>
|
| 75 |
-
📖
|
| 76 |
<br>
|
| 77 |
-
📍
|
| 78 |
<br>
|
| 79 |
-
👉
|
| 80 |
</p>
|
|
|
|
|
|
|
| 81 |
|
| 82 |
-
|
| 83 |
|
| 84 |
-
|
| 85 |
-
**1060** 亿总参数量,其中 **120** 亿活跃参数。GLM-4.5模型统一了推理、编码和智能体能力,以满足智能体应用的复杂需求。
|
| 86 |
|
| 87 |
-
GLM-4.5
|
| 88 |
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
在我们对12项行业标准基准的全面评估中,GLM-4.5表现卓越,得分 **63.2**,在所有专有和开源模型中排名**第3**
|
| 92 |
-
。值得注意的是,GLM-4.5-Air在保持优异效率的同时,仍取得了 **59.8** 的竞争性成绩。
|
| 93 |
|
| 94 |

|
| 95 |
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
|
| 98 |
-
|
| 99 |
-
和 [SGLang](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/glm4_moe.py) 的实现中找到。
|
| 100 |
|
| 101 |
-
|
| 102 |
|
| 103 |
-
请参考我们的[github](https://github.com/zai-org/GLM-4.5)项目。
|
|
|
|
| 7 |
- 量化修复
|
| 8 |
- vLLM
|
| 9 |
base_model:
|
| 10 |
+
- zai-org/GLM-4.5
|
| 11 |
base_model_relation: quantized
|
| 12 |
---
|
| 13 |
# GLM-4.5-AWQ
|
| 14 |
+
Base model [zai-org/GLM-4.5](https://huggingface.co/zai-org/GLM-4.5)
|
| 15 |
|
| 16 |
|
| 17 |
+
### 【VLLM Launch Command for 8-GPU Single Node】
|
| 18 |
+
<i>Note: When launching this model on 8 GPUs, you must include --enable-expert-parallel, otherwise expert tensor partitioning will fail due to mismatch. This flag is not required for 4-GPU setups. </i>
|
| 19 |
```
|
| 20 |
CONTEXT_LENGTH=32768
|
| 21 |
|
| 22 |
vllm serve \
|
| 23 |
+
QuantTrio/GLM-4.5-AWQ \
|
| 24 |
--served-model-name GLM-4.5-AWQ \
|
| 25 |
--enable-expert-parallel \
|
| 26 |
--swap-space 16 \
|
|
|
|
| 35 |
--port 8000
|
| 36 |
```
|
| 37 |
|
| 38 |
+
### 【Dependencies】
|
| 39 |
|
| 40 |
```
|
| 41 |
vllm==0.10.0
|
| 42 |
```
|
| 43 |
|
| 44 |
+
### 【Model Update】
|
| 45 |
```
|
| 46 |
2025-07-30
|
| 47 |
+
1. fast commit
|
| 48 |
```
|
| 49 |
|
| 50 |
+
### 【Model Files】
|
| 51 |
|
| 52 |
+
| File Size | Last Updated |
|
| 53 |
|---------|--------------|
|
| 54 |
| `181GB` | `2025-07-30` |
|
| 55 |
|
| 56 |
|
| 57 |
|
| 58 |
+
### 【Model Download】
|
| 59 |
|
| 60 |
```python
|
| 61 |
+
from huggingface_hub import snapshot_download
|
| 62 |
+
snapshot_download('QuantTrio/GLM-4.5-AWQ', cache_dir="your_local_path")
|
| 63 |
```
|
| 64 |
|
| 65 |
|
| 66 |
+
### 【Overview】
|
| 67 |
# GLM-4.5
|
| 68 |
|
| 69 |
<div align="center">
|
| 70 |
<img src=https://raw.githubusercontent.com/zai-org/GLM-4.5/refs/heads/main/resources/logo.svg width="15%"/>
|
| 71 |
</div>
|
| 72 |
<p align="center">
|
| 73 |
+
👋 Join our <a href="https://discord.gg/QR7SARHRxK" target="_blank">Discord</a> community.
|
| 74 |
<br>
|
| 75 |
+
📖 Check out the GLM-4.5 <a href="https://z.ai/blog/glm-4.5" target="_blank">technical blog</a>.
|
| 76 |
<br>
|
| 77 |
+
📍 Use GLM-4.5 API services on <a href="https://docs.z.ai/guides/llm/glm-4.5">Z.ai API Platform (Global)</a> or <br> <a href="https://docs.bigmodel.cn/cn/guide/models/text/glm-4.5">Zhipu AI Open Platform (Mainland China)</a>.
|
| 78 |
<br>
|
| 79 |
+
👉 One click to <a href="https://chat.z.ai">GLM-4.5</a>.
|
| 80 |
</p>
|
| 81 |
+
|
| 82 |
+
## Model Introduction
|
| 83 |
|
| 84 |
+
The **GLM-4.5** series models are foundation models designed for intelligent agents. GLM-4.5 has **355** billion total parameters with **32** billion active parameters, while GLM-4.5-Air adopts a more compact design with **106** billion total parameters and **12** billion active parameters. GLM-4.5 models unify reasoning, coding, and intelligent agent capabilities to meet the complex demands of intelligent agent applications.
|
| 85 |
|
| 86 |
+
Both GLM-4.5 and GLM-4.5-Air are hybrid reasoning models that provide two modes: thinking mode for complex reasoning and tool usage, and non-thinking mode for immediate responses.
|
|
|
|
| 87 |
|
| 88 |
+
We have open-sourced the base models, hybrid reasoning models, and FP8 versions of the hybrid reasoning models for both GLM-4.5 and GLM-4.5-Air. They are released under the MIT open-source license and can be used commercially and for secondary development.
|
| 89 |
|
| 90 |
+
As demonstrated in our comprehensive evaluation across 12 industry-standard benchmarks, GLM-4.5 achieves exceptional performance with a score of **63.2**, in the **3rd** place among all the proprietary and open-source models. Notably, GLM-4.5-Air delivers competitive results at **59.8** while maintaining superior efficiency.
|
|
|
|
|
|
|
|
|
|
| 91 |
|
| 92 |

|
| 93 |
|
| 94 |
+
For more eval results, show cases, and technical details, please visit
|
| 95 |
+
our [technical blog](https://z.ai/blog/glm-4.5). The technical report will be released soon.
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
The model code, tool parser and reasoning parser can be found in the implementation of [transformers](https://github.com/huggingface/transformers/tree/main/src/transformers/models/glm4_moe), [vLLM](https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/glm4_moe_mtp.py) and [SGLang](https://github.com/sgl-project/sglang/blob/main/python/sglang/srt/models/glm4_moe.py).
|
| 99 |
|
| 100 |
+
## Quick Start
|
|
|
|
| 101 |
|
| 102 |
+
Please refer our [github page](https://github.com/zai-org/GLM-4.5) for more detail.
|
| 103 |
|
|
|