Skip to content

Commit a26c6a1

Browse files
authored
[model] update glm4_5 (#5031)
1 parent fed0147 commit a26c6a1

File tree

5 files changed

+20
-8
lines changed

5 files changed

+20
-8
lines changed

docs/source/Instruction/支持的模型和数据集.md

Lines changed: 6 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -339,8 +339,12 @@
339339
|[ZhipuAI/GLM-4-32B-Base-0414](https://modelscope.cn/models/ZhipuAI/GLM-4-32B-Base-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-4-32B-Base-0414](https://huggingface.co/THUDM/GLM-4-32B-Base-0414)|
340340
|[ZhipuAI/GLM-Z1-9B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-9B-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-Z1-9B-0414](https://huggingface.co/THUDM/GLM-Z1-9B-0414)|
341341
|[ZhipuAI/GLM-Z1-32B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-32B-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-Z1-32B-0414](https://huggingface.co/THUDM/GLM-Z1-32B-0414)|
342-
|[ZhipuAI/GLM-4.5-MOE-106B-A12B-0715](https://modelscope.cn/models/ZhipuAI/GLM-4.5-MOE-106B-A12B-0715)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-MOE-106B-A12B-0715](https://huggingface.co/THUDM/GLM-4.5-MOE-106B-A12B-0715)|
343-
|[ZhipuAI/GLM-4.5-MOE-355B-A32B-0715](https://modelscope.cn/models/ZhipuAI/GLM-4.5-MOE-355B-A32B-0715)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-MOE-355B-A32B-0715](https://huggingface.co/THUDM/GLM-4.5-MOE-355B-A32B-0715)|
342+
|[ZhipuAI/GLM-4.5-Air-Base](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air-Base)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Air-Base](https://huggingface.co/THUDM/GLM-4.5-Air-Base)|
343+
|[ZhipuAI/GLM-4.5-Air](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Air](https://huggingface.co/THUDM/GLM-4.5-Air)|
344+
|[ZhipuAI/GLM-4.5-Air-FP8](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air-FP8)|glm4_5|glm4_5|transformers>=4.54|✘|-|[THUDM/GLM-4.5-Air-FP8](https://huggingface.co/THUDM/GLM-4.5-Air-FP8)|
345+
|[ZhipuAI/GLM-4.5-Base](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Base)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Base](https://huggingface.co/THUDM/GLM-4.5-Base)|
346+
|[ZhipuAI/GLM-4.5](https://modelscope.cn/models/ZhipuAI/GLM-4.5)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5](https://huggingface.co/THUDM/GLM-4.5)|
347+
|[ZhipuAI/GLM-4.5-FP8](https://modelscope.cn/models/ZhipuAI/GLM-4.5-FP8)|glm4_5|glm4_5|transformers>=4.54|✘|-|[THUDM/GLM-4.5-FP8](https://huggingface.co/THUDM/GLM-4.5-FP8)|
344348
|[ZhipuAI/GLM-Z1-Rumination-32B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-Rumination-32B-0414)|glm4_z1_rumination|glm4_z1_rumination|transformers>4.51|✘|-|[THUDM/GLM-Z1-Rumination-32B-0414](https://huggingface.co/THUDM/GLM-Z1-Rumination-32B-0414)|
345349
|[ZhipuAI/glm-edge-1.5b-chat](https://modelscope.cn/models/ZhipuAI/glm-edge-1.5b-chat)|glm_edge|glm4|transformers>=4.46|✘|-|[THUDM/glm-edge-1.5b-chat](https://huggingface.co/THUDM/glm-edge-1.5b-chat)|
346350
|[ZhipuAI/glm-edge-4b-chat](https://modelscope.cn/models/ZhipuAI/glm-edge-4b-chat)|glm_edge|glm4|transformers>=4.46|✘|-|[THUDM/glm-edge-4b-chat](https://huggingface.co/THUDM/glm-edge-4b-chat)|

docs/source_en/Instruction/Supported-models-and-datasets.md

Lines changed: 6 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -339,8 +339,12 @@ The table below introduces the models integrated with ms-swift:
339339
|[ZhipuAI/GLM-4-32B-Base-0414](https://modelscope.cn/models/ZhipuAI/GLM-4-32B-Base-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-4-32B-Base-0414](https://huggingface.co/THUDM/GLM-4-32B-Base-0414)|
340340
|[ZhipuAI/GLM-Z1-9B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-9B-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-Z1-9B-0414](https://huggingface.co/THUDM/GLM-Z1-9B-0414)|
341341
|[ZhipuAI/GLM-Z1-32B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-32B-0414)|glm4_0414|glm4_0414|transformers>=4.51|✘|-|[THUDM/GLM-Z1-32B-0414](https://huggingface.co/THUDM/GLM-Z1-32B-0414)|
342-
|[ZhipuAI/GLM-4.5-MOE-106B-A12B-0715](https://modelscope.cn/models/ZhipuAI/GLM-4.5-MOE-106B-A12B-0715)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-MOE-106B-A12B-0715](https://huggingface.co/THUDM/GLM-4.5-MOE-106B-A12B-0715)|
343-
|[ZhipuAI/GLM-4.5-MOE-355B-A32B-0715](https://modelscope.cn/models/ZhipuAI/GLM-4.5-MOE-355B-A32B-0715)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-MOE-355B-A32B-0715](https://huggingface.co/THUDM/GLM-4.5-MOE-355B-A32B-0715)|
342+
|[ZhipuAI/GLM-4.5-Air-Base](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air-Base)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Air-Base](https://huggingface.co/THUDM/GLM-4.5-Air-Base)|
343+
|[ZhipuAI/GLM-4.5-Air](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Air](https://huggingface.co/THUDM/GLM-4.5-Air)|
344+
|[ZhipuAI/GLM-4.5-Air-FP8](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Air-FP8)|glm4_5|glm4_5|transformers>=4.54|✘|-|[THUDM/GLM-4.5-Air-FP8](https://huggingface.co/THUDM/GLM-4.5-Air-FP8)|
345+
|[ZhipuAI/GLM-4.5-Base](https://modelscope.cn/models/ZhipuAI/GLM-4.5-Base)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5-Base](https://huggingface.co/THUDM/GLM-4.5-Base)|
346+
|[ZhipuAI/GLM-4.5](https://modelscope.cn/models/ZhipuAI/GLM-4.5)|glm4_5|glm4_5|transformers>=4.54|✔|-|[THUDM/GLM-4.5](https://huggingface.co/THUDM/GLM-4.5)|
347+
|[ZhipuAI/GLM-4.5-FP8](https://modelscope.cn/models/ZhipuAI/GLM-4.5-FP8)|glm4_5|glm4_5|transformers>=4.54|✘|-|[THUDM/GLM-4.5-FP8](https://huggingface.co/THUDM/GLM-4.5-FP8)|
344348
|[ZhipuAI/GLM-Z1-Rumination-32B-0414](https://modelscope.cn/models/ZhipuAI/GLM-Z1-Rumination-32B-0414)|glm4_z1_rumination|glm4_z1_rumination|transformers>4.51|✘|-|[THUDM/GLM-Z1-Rumination-32B-0414](https://huggingface.co/THUDM/GLM-Z1-Rumination-32B-0414)|
345349
|[ZhipuAI/glm-edge-1.5b-chat](https://modelscope.cn/models/ZhipuAI/glm-edge-1.5b-chat)|glm_edge|glm4|transformers>=4.46|✘|-|[THUDM/glm-edge-1.5b-chat](https://huggingface.co/THUDM/glm-edge-1.5b-chat)|
346350
|[ZhipuAI/glm-edge-4b-chat](https://modelscope.cn/models/ZhipuAI/glm-edge-4b-chat)|glm_edge|glm4|transformers>=4.46|✘|-|[THUDM/glm-edge-4b-chat](https://huggingface.co/THUDM/glm-edge-4b-chat)|

swift/llm/model/model/glm.py

Lines changed: 6 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -425,8 +425,12 @@ def get_model_tokenizer_glm_edge_v(model_dir: str, *args, **kwargs):
425425
LLMModelType.glm4_5,
426426
[
427427
ModelGroup([
428-
Model('ZhipuAI/GLM-4.5-MOE-106B-A12B-0715', 'THUDM/GLM-4.5-MOE-106B-A12B-0715'),
429-
Model('ZhipuAI/GLM-4.5-MOE-355B-A32B-0715', 'THUDM/GLM-4.5-MOE-355B-A32B-0715'),
428+
Model('ZhipuAI/GLM-4.5-Air-Base', 'THUDM/GLM-4.5-Air-Base'),
429+
Model('ZhipuAI/GLM-4.5-Air', 'THUDM/GLM-4.5-Air'),
430+
Model('ZhipuAI/GLM-4.5-Air-FP8', 'THUDM/GLM-4.5-Air-FP8'),
431+
Model('ZhipuAI/GLM-4.5-Base', 'THUDM/GLM-4.5-Base'),
432+
Model('ZhipuAI/GLM-4.5', 'THUDM/GLM-4.5'),
433+
Model('ZhipuAI/GLM-4.5-FP8', 'THUDM/GLM-4.5-FP8'),
430434
]),
431435
],
432436
TemplateType.glm4_5,

tests/megatron/test_align/test_llm.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -124,7 +124,7 @@ def test_ernie():
124124

125125

126126
def test_glm4_5():
127-
_test_model('ZhipuAI/GLM-4.5-MOE-106B-A12B-0715')
127+
_test_model('ZhipuAI/GLM-4.5-Air')
128128

129129

130130
if __name__ == '__main__':

tests/test_align/test_template/test_llm.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -448,7 +448,7 @@ def test_ernie():
448448

449449
def test_glm4_5():
450450
messages = [{'role': 'user', 'content': '浙江的省会在哪?'}]
451-
pt_engine = PtEngine('ZhipuAI/GLM-4.5-MOE-106B-A12B-0715')
451+
pt_engine = PtEngine('ZhipuAI/GLM-4.5-Air')
452452
res = _infer_model(pt_engine, messages=messages)
453453
pt_engine.default_template.template_backend = 'jinja'
454454
res2 = _infer_model(pt_engine, messages=messages)

0 commit comments

Comments
 (0)