A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 12:15 UTC
Remove Bamba 9B from CI (#17407)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Harry Mellor committed
Apr 29, 2025 at 22:10 UTC
7489ec0bab2904dcc4001af59a942a16756fdbbc
3 files changed
+4
-3
tests/models/decoder_only/language/test_hybrid.py
+1
-1
@@ -28,7 +28,7 @@ HYBRID_MODELS = [
28
# not compatible with pip-compile.
29
"pfnet/plamo-2-1b",
30
"Zyphra/Zamba2-1.2B-instruct",
31
- "ibm-ai-platform/Bamba-9B",
31
+ "hmellor/bamba-tiny-random",
32
]
33
34
# Avoid OOM
tests/models/registry.py
+2
-1
@@ -120,7 +120,8 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
120
trust_remote_code=True),
121
"BaichuanForCausalLM": _HfExamplesInfo("baichuan-inc/Baichuan2-7B-chat",
122
trust_remote_code=True),
123
- "BambaForCausalLM": _HfExamplesInfo("ibm-ai-platform/Bamba-9B"),
123
+ "BambaForCausalLM": _HfExamplesInfo("ibm-ai-platform/Bamba-9B",
124
+ extras={"tiny": "hmellor/bamba-tiny-random"}), # noqa: E501
125
"BloomForCausalLM": _HfExamplesInfo("bigscience/bloom-560m",
126
{"1b": "bigscience/bloomz-1b1"}),
127
"ChatGLMModel": _HfExamplesInfo("THUDM/chatglm3-6b",
tests/v1/test_oracle.py
+1
-1
@@ -12,7 +12,7 @@ UNSUPPORTED_MODELS_V1 = [
12
"openai/whisper-large-v3", # transcription
13
"facebook/bart-large-cnn", # encoder decoder
14
"mistralai/Mamba-Codestral-7B-v0.1", # mamba
15
- "ibm-ai-platform/Bamba-9B", # hybrid
15
+ "hmellor/bamba-tiny-random", # hybrid
16
"BAAI/bge-m3", # embedding
17
]
18