[v0.11.0-dev][Bugfix][cherry-pick]bugfix for weight load of kimi-k2 (vllm-project#4190)

### What this PR does / why we need it?
This is cherry-pick from vllm-project#3798 

Fix kimi-k2 start bug, weight load
ERROR：vllm-project#3785

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.11.0rc3
- vLLM main:
vllm-project/vllm@c9461e0

---------

Signed-off-by: Levi-JQ <[email protected]>
Signed-off-by: menogrey <[email protected]>
Co-authored-by: Levi <[email protected]>
Co-authored-by: Levi-JQ <[email protected]>
Co-authored-by: zhaozx-cn <[email protected]>

Author: 4 people

.github/workflows/release_whl.yml

@@ -57,7 +57,13 @@ jobs:
     - name: Print
       run: |
         lscpu
-        
+
+    - name: Free up disk space
+      uses: jlumbroso/free-disk-space@54081f138730dfa15788a46383842cd2f914a1be # v1.3.1
+      with:
+        tool-cache: true
+        docker-images: false
+
     - name: Build wheel
       run: |
         ls

vllm_ascend/quantization/quant_config.py

@@ -193,6 +193,11 @@ def get_scaled_act_names(self) -> List[str]:
         ["experts.0.gate_proj", "experts.0.up_proj", "experts.0.down_proj"],
         "fused_qkv_a_proj": ["q_a_proj", "kv_a_proj_with_mqa"]
     },
+    "kimi_k2": {
+        "gate_up_proj": ["gate_proj", "up_proj"],
+        "experts":
+        ["experts.0.gate_proj", "experts.0.up_proj", "experts.0.down_proj"]
+    },
     "deepseek_v32": {
         "gate_up_proj": ["gate_proj", "up_proj"],
         "experts":