[2/N][Refactor][Qwen3-Next] remove redundant methods and patch methods in Qwen3NextGatedDeltaNet (#3082)
### What this PR does / why we need it?
remove redundant methods and patch methods in Qwen3NextGatedDeltaNet
involved causal_conv1d_fn, causal_conv1d_update_npu, fused_gdn_gating,
fused_reccrrent_gated_delta_rule, torch_chunk_gated_delta_rule,
RMSNormGated
### Does this PR introduce _any_ user-facing change?
N/A
### How was this patch tested?
```
def main():
prompts = [
"The future of AI is",
]
# Create a sampling params object.
sampling_params = SamplingParams(max_tokens=100, temperature=0.6, top_k=40, top_p=0.95)
# Create an LLM.
llm = LLM(
model="Qwen/Qwen3-Next-80B-A3B-Instruct",
tensor_parallel_size=4,
enforce_eager=True,
trust_remote_code=True,
max_model_len=256,
gpu_memory_utilization=0.7,
block_size=64,
)
# Generate texts from the prompts.
outputs = llm.generate(prompts, sampling_params)
for output in outputs:
prompt = output.prompt
generated_text = output.outputs[0].text
print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```
CI passed with new added/existing test.
- vLLM version: v0.10.2
- vLLM main:
5aeb925452
---------
Signed-off-by: Icey <1790571317@qq.com>
This commit is contained in:
@@ -15,6 +15,11 @@
|
||||
# limitations under the License.
|
||||
#
|
||||
|
||||
from vllm.triton_utils import HAS_TRITON
|
||||
|
||||
if HAS_TRITON:
|
||||
import vllm_ascend.patch.worker.patch_common.patch_triton
|
||||
|
||||
import vllm_ascend.patch.worker.patch_common.patch_distributed # noqa
|
||||
import vllm_ascend.patch.worker.patch_common.patch_logits # noqa
|
||||
|
||||
|
||||
16
vllm_ascend/patch/worker/patch_common/patch_triton.py
Normal file
16
vllm_ascend/patch/worker/patch_common/patch_triton.py
Normal file
@@ -0,0 +1,16 @@
|
||||
import vllm.model_executor.layers.fla.ops.chunk
|
||||
import vllm.model_executor.layers.fla.ops.fused_recurrent
|
||||
import vllm.model_executor.layers.fla.ops.layernorm_guard
|
||||
import vllm.model_executor.layers.mamba.ops.causal_conv1d
|
||||
|
||||
from vllm_ascend.ops.casual_conv1d import (causal_conv1d_fn,
|
||||
causal_conv1d_update_npu)
|
||||
from vllm_ascend.ops.fla import LayerNormFn, torch_chunk_gated_delta_rule
|
||||
from vllm_ascend.ops.sigmoid_gating import \
|
||||
fused_recurrent_gated_delta_rule_fwd_kernel
|
||||
|
||||
vllm.model_executor.layers.mamba.ops.causal_conv1d.causal_conv1d_update = causal_conv1d_update_npu
|
||||
vllm.model_executor.layers.mamba.ops.causal_conv1d.causal_conv1d_fn = causal_conv1d_fn
|
||||
vllm.model_executor.layers.fla.ops.fused_recurrent.fused_recurrent_gated_delta_rule_fwd_kernel = fused_recurrent_gated_delta_rule_fwd_kernel
|
||||
vllm.model_executor.layers.fla.ops.layernorm_guard.LayerNormFn = LayerNormFn
|
||||
vllm.model_executor.layers.fla.ops.chunk.chunk_gated_delta_rule = torch_chunk_gated_delta_rule
|
||||
Reference in New Issue
Block a user