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

[MM][Bugfix] Update hf_config to hf_text_config (#5319)

Browse Source 
### What this PR does / why we need it?

Following https://github.com/vllm-project/vllm-ascend/pull/5205, update
`hf_config` to `hf_text_config`.

Find more details at
https://github.com/vllm-project/vllm-ascend/pull/5205#issuecomment-3675417534
and
https://github.com/vllm-project/vllm-ascend/pull/5205#issuecomment-3677920872.

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

### How was this patch tested?

- vLLM version: release/v0.13.0
- vLLM main:
5fbfa8d9ef

Signed-off-by: shen-shanshan <467638484@qq.com>
This commit is contained in:
Shanshan Shen
2026-01-06 16:41:39 +08:00
committed by GitHub
parent 293b2275df
commit b94d589769
23 changed files with 44 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm_ascend/worker/model_runner_v1.py
View File

@@ -253,7 +253,7 @@ class NPUModelRunner(GPUModelRunner):
self.is_multimodal_model = self.model_config.is_multimodal_model
self.block_size = vllm_config.cache_config.block_size
# Set up Attention
self.use_sparse = hasattr(self.vllm_config.model_config.hf_config,
self.use_sparse = hasattr(self.vllm_config.model_config.hf_text_config,
"index_topk")
self.attn_backend = get_attn_backend(
0,
@@ -2398,7 +2398,7 @@ class NPUModelRunner(GPUModelRunner):
kv_caches[layer_name] = kv_caches[target_layer_name]
from vllm.v1.worker.utils import bind_kv_cache
num_attn_module = 2 if self.model_config.hf_config.model_type == "longcat_flash" else 1
num_attn_module = 2 if self.model_config.hf_text_config.model_type == "longcat_flash" else 1
bind_kv_cache(kv_caches,
self.compilation_config.static_forward_context,
self.kv_caches, num_attn_module)
@@ -2932,7 +2932,7 @@ class NPUModelRunner(GPUModelRunner):
mamba_layers = get_layers_from_vllm_config(self.vllm_config, MambaBase)
if len(mamba_layers) > 0:
if (self.vllm_config.speculative_config is not None
and self.vllm_config.model_config.hf_config.model_type
and self.vllm_config.model_config.hf_text_config.model_type
not in ["qwen3_next"]):
raise NotImplementedError(
"Mamba with speculative decoding is not supported yet.")

2
vllm_ascend/worker/worker.py
View File

@@ -173,7 +173,7 @@ class NPUWorker(WorkerBase):
allocator = CaMemAllocator.get_instance()
allocator.wake_up(tags=tags)
hidden_size = self.vllm_config.model_config.hf_config.hidden_size
hidden_size = self.vllm_config.model_config.hf_text_config.hidden_size
model = self.model_runner.model
if tags is None or "weights" in tags:
for name, param in model.named_parameters():