EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[2/N][Pangu][MoE] Remove Pangu Related Code (#5130)

Browse Source 
### What this PR does / why we need it?
Remove Pangu Related Code

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
e2e & ut

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: weichen <calvin_zhu0210@outlook.com>
This commit is contained in:
weichen
2025-12-19 09:00:07 +08:00
committed by GitHub
parent 1b47fca0e8
commit ca6f631cba
11 changed files with 8 additions and 1444 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
vllm_ascend/worker/model_runner_v1.py
View File

@@ -2504,14 +2504,8 @@ class NPUModelRunner(GPUModelRunner):
# the min of all `num_blocks`. Verify it here.
assert num_blocks >= kv_cache_config.num_blocks
if self.vllm_config.additional_config.get(
"kv_cache_dtype", None) == 'int8':
kv_cache_shape = attn_backend.get_bsh_kv_cache_shape(
num_blocks, kv_cache_spec.block_size,
kv_cache_spec.num_kv_heads,
kv_cache_spec.head_size)
elif hasattr(attn_backend, "get_supported_block_size"
) and self.use_hybrid_blocks:
if hasattr(attn_backend, "get_supported_block_size"
) and self.use_hybrid_blocks:
block_size = attn_backend.get_supported_block_size()[0]
block_size_chunk = kv_cache_spec.block_size // block_size