Unverified Commit 21eb2c33 authored by Isotr0py's avatar Isotr0py Committed by GitHub
Browse files

[Chore] Correct MTP models test registry ordering (#36115)


Signed-off-by: default avatarIsotr0py <mozf@mail2.sysu.edu.cn>
parent e2b31243
...@@ -421,11 +421,6 @@ _TEXT_GENERATION_EXAMPLE_MODELS = { ...@@ -421,11 +421,6 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
"Olmo2ForCausalLM": _HfExamplesInfo("allenai/OLMo-2-0425-1B"), "Olmo2ForCausalLM": _HfExamplesInfo("allenai/OLMo-2-0425-1B"),
"Olmo3ForCausalLM": _HfExamplesInfo("allenai/Olmo-3-7B-Instruct"), "Olmo3ForCausalLM": _HfExamplesInfo("allenai/Olmo-3-7B-Instruct"),
"OlmoeForCausalLM": _HfExamplesInfo("allenai/OLMoE-1B-7B-0924-Instruct"), "OlmoeForCausalLM": _HfExamplesInfo("allenai/OLMoE-1B-7B-0924-Instruct"),
"OpenPanguMTPModel": _HfExamplesInfo(
"FreedomIntelligence/openPangu-Ultra-MoE-718B-V1.1",
trust_remote_code=True,
is_available_online=False,
),
"OPTForCausalLM": _HfExamplesInfo( "OPTForCausalLM": _HfExamplesInfo(
"facebook/opt-125m", {"1b": "facebook/opt-iml-max-1.3b"} "facebook/opt-125m", {"1b": "facebook/opt-iml-max-1.3b"}
), ),
...@@ -1018,14 +1013,6 @@ _MULTIMODAL_EXAMPLE_MODELS = { ...@@ -1018,14 +1013,6 @@ _MULTIMODAL_EXAMPLE_MODELS = {
"Qwen/Qwen3.5-35B-A3B", "Qwen/Qwen3.5-35B-A3B",
max_model_len=4096, max_model_len=4096,
), ),
"Qwen3_5MTP": _HfExamplesInfo(
"Qwen/Qwen3.5-0.8B",
speculative_model="Qwen/Qwen3.5-0.8B",
),
"Qwen3_5MoeMTP": _HfExamplesInfo(
"Qwen/Qwen3.5-35B-A3B",
speculative_model="Qwen/Qwen3.5-35B-A3B",
),
"Qwen3OmniMoeForConditionalGeneration": _HfExamplesInfo( "Qwen3OmniMoeForConditionalGeneration": _HfExamplesInfo(
"Qwen/Qwen3-Omni-30B-A3B-Instruct", "Qwen/Qwen3-Omni-30B-A3B-Instruct",
max_model_len=4096, max_model_len=4096,
...@@ -1093,6 +1080,7 @@ _MULTIMODAL_EXAMPLE_MODELS = { ...@@ -1093,6 +1080,7 @@ _MULTIMODAL_EXAMPLE_MODELS = {
_SPECULATIVE_DECODING_EXAMPLE_MODELS = { _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
# [Medusa]
"MedusaModel": _HfExamplesInfo( "MedusaModel": _HfExamplesInfo(
"JackFram/llama-68m", speculative_model="abhigoyal/vllm-medusa-llama-68m-random" "JackFram/llama-68m", speculative_model="abhigoyal/vllm-medusa-llama-68m-random"
), ),
...@@ -1102,11 +1090,7 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = { ...@@ -1102,11 +1090,7 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
# "JackFram/llama-160m", # "JackFram/llama-160m",
# speculative_model="ibm-ai-platform/llama-160m-accelerator" # speculative_model="ibm-ai-platform/llama-160m-accelerator"
# ), # ),
"DeepSeekMTPModel": _HfExamplesInfo( # [Eagle]
"luccafong/deepseek_mtp_main_random",
speculative_model="luccafong/deepseek_mtp_draft_random",
trust_remote_code=True,
),
"EagleDeepSeekMTPModel": _HfExamplesInfo( "EagleDeepSeekMTPModel": _HfExamplesInfo(
"eagle618/deepseek-v3-random", "eagle618/deepseek-v3-random",
speculative_model="eagle618/eagle-deepseek-v3-random", speculative_model="eagle618/eagle-deepseek-v3-random",
...@@ -1152,6 +1136,20 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = { ...@@ -1152,6 +1136,20 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
speculative_method="eagle", speculative_method="eagle",
tokenizer="openbmb/MiniCPM-2B-sft-bf16", tokenizer="openbmb/MiniCPM-2B-sft-bf16",
), ),
"Eagle3Qwen2_5vlForCausalLM": _HfExamplesInfo(
"Qwen/Qwen2.5-VL-7B-Instruct",
speculative_model="Rayzl/qwen2.5-vl-7b-eagle3-sgl",
),
"Eagle3Qwen3vlForCausalLM": _HfExamplesInfo(
"Qwen/Qwen3-VL-8B-Instruct",
speculative_model="taobao-mnn/Qwen3-VL-8B-Instruct-Eagle3",
),
# [MTP]
"DeepSeekMTPModel": _HfExamplesInfo(
"luccafong/deepseek_mtp_main_random",
speculative_model="luccafong/deepseek_mtp_draft_random",
trust_remote_code=True,
),
"ErnieMTPModel": _HfExamplesInfo( "ErnieMTPModel": _HfExamplesInfo(
"baidu/ERNIE-4.5-21B-A3B-PT", "baidu/ERNIE-4.5-21B-A3B-PT",
trust_remote_code=True, trust_remote_code=True,
...@@ -1191,17 +1189,27 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = { ...@@ -1191,17 +1189,27 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
trust_remote_code=True, trust_remote_code=True,
speculative_model="XiaomiMiMo/MiMo-7B-RL", speculative_model="XiaomiMiMo/MiMo-7B-RL",
), ),
"Eagle3Qwen2_5vlForCausalLM": _HfExamplesInfo( "NemotronHMTPModel": _HfExamplesInfo(
"Qwen/Qwen2.5-VL-7B-Instruct", "nvidia/Nemotron-Super-Placeholder",
speculative_model="Rayzl/qwen2.5-vl-7b-eagle3-sgl", speculative_model="nvidia/Nemotron-Super-Placeholder",
is_available_online=False,
), ),
"Eagle3Qwen3vlForCausalLM": _HfExamplesInfo( "OpenPanguMTPModel": _HfExamplesInfo(
"Qwen/Qwen3-VL-8B-Instruct", "FreedomIntelligence/openPangu-Ultra-MoE-718B-V1.1",
speculative_model="taobao-mnn/Qwen3-VL-8B-Instruct-Eagle3", trust_remote_code=True,
is_available_online=False,
), ),
"Qwen3NextMTP": _HfExamplesInfo( "Qwen3NextMTP": _HfExamplesInfo(
"Qwen/Qwen3-Next-80B-A3B-Instruct", min_transformers_version="4.56.3" "Qwen/Qwen3-Next-80B-A3B-Instruct", min_transformers_version="4.56.3"
), ),
"Qwen3_5MTP": _HfExamplesInfo(
"Qwen/Qwen3.5-0.8B",
speculative_model="Qwen/Qwen3.5-0.8B",
),
"Qwen3_5MoeMTP": _HfExamplesInfo(
"Qwen/Qwen3.5-35B-A3B",
speculative_model="Qwen/Qwen3.5-35B-A3B",
),
"Step3p5MTP": _HfExamplesInfo( "Step3p5MTP": _HfExamplesInfo(
"stepfun-ai/Step-3.5-Flash", "stepfun-ai/Step-3.5-Flash",
speculative_model="stepfun-ai/Step-3.5-Flash", speculative_model="stepfun-ai/Step-3.5-Flash",
...@@ -1212,11 +1220,6 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = { ...@@ -1212,11 +1220,6 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
}, },
is_available_online=False, is_available_online=False,
), ),
"NemotronHMTPModel": _HfExamplesInfo(
"nvidia/Nemotron-Super-Placeholder",
speculative_model="nvidia/Nemotron-Super-Placeholder",
is_available_online=False,
),
} }
_TRANSFORMERS_BACKEND_MODELS = { _TRANSFORMERS_BACKEND_MODELS = {
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment