xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 08:14:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

self.gate dtype update for GLM-4.5 (#22203)

Browse Source 
Signed-off-by: zRzRzRzRzRzRzR <2448370773@qq.com>
This commit is contained in:
Yuxuan Zhang 2025-08-05 10:12:38 +08:00 committed by GitHub
parent 031ca762d7
commit 6fa41e0c32
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 4 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/models/supported_models.md
View File

@ -606,7 +606,7 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
| `GLM4VForCausalLM`<sup>^</sup> | GLM-4V | T + I | `zai-org/glm-4v-9b`, `zai-org/cogagent-9b-20241220`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4vForConditionalGeneration` | GLM-4.1V-Thinking | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.1V-9B-Thinking`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4MoeForCausalLM` | GLM-4.5 | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4v_moeForConditionalGeneration` | GLM-4.5V | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5V-Air`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4v_moeForConditionalGeneration` | GLM-4.5V | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5V`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GraniteSpeechForConditionalGeneration` | Granite Speech | T + A | `ibm-granite/granite-speech-3.3-8b` | ✅︎ | ✅︎ | ✅︎ |
| `H2OVLChatModel` | H2OVL | T + I<sup>E+</sup> | `h2oai/h2ovl-mississippi-800m`, `h2oai/h2ovl-mississippi-2b`, etc. | | ✅︎ | ✅︎ |
| `Idefics3ForConditionalGeneration` | Idefics3 | T + I | `HuggingFaceM4/Idefics3-8B-Llama3`, etc. | ✅︎ | | ✅︎ |

2
tests/models/registry.py
View File

@ -383,7 +383,7 @@ _MULTIMODAL_EXAMPLE_MODELS = {
trust_remote_code=True,
hf_overrides={"architectures": ["GLM4VForCausalLM"]}), # noqa: E501
"Glm4vForConditionalGeneration": _HfExamplesInfo("zai-org/GLM-4.1V-9B-Thinking"), # noqa: E501
"Glm4v_moeForConditionalGeneration": _HfExamplesInfo("zai-org/GLM-4.5V-Air",
"Glm4v_moeForConditionalGeneration": _HfExamplesInfo("zai-org/GLM-4.5V",
is_available_online=False), # noqa: E501
"H2OVLChatModel": _HfExamplesInfo("h2oai/h2ovl-mississippi-800m",
trust_remote_code=True,

3
vllm/model_executor/models/glm4_moe.py
View File

@ -123,6 +123,7 @@ class Glm4MoE(nn.Module):
config.n_routed_experts,
bias=False,
quant_config=None,
params_dtype=torch.float32,
prefix=f"{prefix}.gate")
self.gate.e_score_correction_bias = nn.Parameter(
@ -180,7 +181,7 @@ class Glm4MoE(nn.Module):
if self.n_shared_experts is not None:
shared_output = self.shared_experts(hidden_states)
router_logits, _ = self.gate(hidden_states)
router_logits, _ = self.gate(hidden_states.to(dtype=torch.float32))
final_hidden_states = self.experts(
hidden_states=hidden_states,
router_logits=router_logits) * self.routed_scaling_factor