xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-06 22:35:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add SSM and Hybrid Models Test

Browse Source 
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
This commit is contained in:
Tyler Michael Smith 2025-03-15 17:26:01 +00:00
parent 61c6a5a796
commit 696245c2fc
2 changed files with 571 additions and 566 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
.buildkite/test-pipeline.yaml
View File

@ -390,6 +390,15 @@ steps:
- pytest -v -s models/decoder_only/language -m 'core_model or quant_model' - pytest -v -s models/decoder_only/language -m 'core_model or quant_model'
- pytest -v -s models/embedding/language -m core_model - pytest -v -s models/embedding/language -m core_model
- label: SSM and Hybrid Models Test # X min
source_file_dependencies:
- vllm/
- tests/models/decoder_only/language/test_hybrid.py
- tests/models/decoder_only/language/test_mamba.py
commands:
- pytest -v -s models/decoder_only/language/test_hybrid.py
- pytest -v -s models/decoder_only/language/test_mamba.py
- label: Language Models Test (Extended) # 1h10min - label: Language Models Test (Extended) # 1h10min
optional: true optional: true
source_file_dependencies: source_file_dependencies:
@ -593,7 +602,6 @@ steps:
- pytest -v -s -x lora/test_minicpmv_tp.py - pytest -v -s -x lora/test_minicpmv_tp.py
- pytest -v -s -x lora/test_transfomers_model.py - pytest -v -s -x lora/test_transfomers_model.py
- label: Weight Loading Multiple GPU Test # 33min - label: Weight Loading Multiple GPU Test # 33min
working_dir: "/vllm-workspace/tests" working_dir: "/vllm-workspace/tests"
num_gpus: 2 num_gpus: 2
@ -614,7 +622,6 @@ steps:
commands: commands:
- bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models-large.txt - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models-large.txt
##### multi gpus test ##### ##### multi gpus test #####
##### A100 test ##### ##### A100 test #####

2
vllm/model_executor/models/bamba.py
View File

@ -38,8 +38,6 @@ from .utils import (is_pp_missing_parameter,
make_empty_intermediate_tensors_factory, make_layers, make_empty_intermediate_tensors_factory, make_layers,
maybe_prefix) maybe_prefix)
KVCache = Tuple[torch.Tensor, torch.Tensor]
class BambaMLP(nn.Module): class BambaMLP(nn.Module):