From 656c24f1b5d8a662e9ec391503d146341def5f18 Mon Sep 17 00:00:00 2001 From: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com> Date: Mon, 28 Jul 2025 14:22:32 +0200 Subject: [PATCH] [`Ernie 4.5`] Name Change for Base 0.3B Model (#21735) Signed-off-by: vasqu Signed-off-by: DarkLight1337 Co-authored-by: DarkLight1337 --- docs/models/supported_models.md | 10 +++++----- tests/models/registry.py | 2 +- vllm/model_executor/models/ernie45.py | 2 +- vllm/model_executor/models/registry.py | 2 +- 4 files changed, 8 insertions(+), 8 deletions(-) diff --git a/docs/models/supported_models.md b/docs/models/supported_models.md index 4bc8858baef16..763dd8fd50451 100644 --- a/docs/models/supported_models.md +++ b/docs/models/supported_models.md @@ -336,7 +336,7 @@ th { | `DeepseekV2ForCausalLM` | DeepSeek-V2 | `deepseek-ai/DeepSeek-V2`, `deepseek-ai/DeepSeek-V2-Chat`, etc. | | ✅︎ | ✅︎ | | `DeepseekV3ForCausalLM` | DeepSeek-V3 | `deepseek-ai/DeepSeek-V3-Base`, `deepseek-ai/DeepSeek-V3`, etc. | | ✅︎ | ✅︎ | | `Dots1ForCausalLM` | dots.llm1 | `rednote-hilab/dots.llm1.base`, `rednote-hilab/dots.llm1.inst`, etc. | | ✅︎ | ✅︎ | -| `Ernie4_5_ForCausalLM` | Ernie4.5 | `baidu/ERNIE-4.5-0.3B-PT`, etc. | ✅︎ | ✅︎ | ✅︎ | +| `Ernie4_5ForCausalLM` | Ernie4.5 | `baidu/ERNIE-4.5-0.3B-PT`, etc. | ✅︎ | ✅︎ | ✅︎ | | `Ernie4_5_MoeForCausalLM` | Ernie4.5MoE | `baidu/ERNIE-4.5-21B-A3B-PT`, `baidu/ERNIE-4.5-300B-A47B-PT`, etc. |✅︎| ✅︎ | ✅︎ | | `ExaoneForCausalLM` | EXAONE-3 | `LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct`, etc. | ✅︎ | ✅︎ | ✅︎ | | `Exaone4ForCausalLM` | EXAONE-4 | `LGAI-EXAONE/EXAONE-4.0-32B`, etc. | ✅︎ | ✅︎ | ✅︎ | @@ -634,10 +634,10 @@ Some models are supported only via the [Transformers backend](#transformers). Th |--------------|--------|--------|-------------------|-----------------------------|-----------------------------------------|---------------------| | `Emu3ForConditionalGeneration` | Emu3 | T + I | `BAAI/Emu3-Chat-hf` | ✅︎ | ✅︎ | ✅︎ | -^ You need to set the architecture name via `--hf-overrides` to match the one in vLLM. -    • For example, to use DeepSeek-VL2 series models: -      `--hf-overrides '{"architectures": ["DeepseekVLV2ForCausalLM"]}'` -E Pre-computed embeddings can be inputted for this modality. +^ You need to set the architecture name via `--hf-overrides` to match the one in vLLM. +    • For example, to use DeepSeek-VL2 series models: +      `--hf-overrides '{"architectures": ["DeepseekVLV2ForCausalLM"]}'` +E Pre-computed embeddings can be inputted for this modality. + Multiple items can be inputted per text prompt for this modality. !!! warning diff --git a/tests/models/registry.py b/tests/models/registry.py index 0ef2a028b4a1e..4fcd02efb6d0b 100644 --- a/tests/models/registry.py +++ b/tests/models/registry.py @@ -166,7 +166,7 @@ _TEXT_GENERATION_EXAMPLE_MODELS = { trust_remote_code=True), "DeepseekV3ForCausalLM": _HfExamplesInfo("deepseek-ai/DeepSeek-V3", # noqa: E501 trust_remote_code=True), - "Ernie4_5_ForCausalLM": _HfExamplesInfo("baidu/ERNIE-4.5-0.3B-PT", + "Ernie4_5ForCausalLM": _HfExamplesInfo("baidu/ERNIE-4.5-0.3B-PT", min_transformers_version="4.54"), "Ernie4_5_MoeForCausalLM": _HfExamplesInfo("baidu/ERNIE-4.5-21B-A3B-PT", min_transformers_version="4.54"), diff --git a/vllm/model_executor/models/ernie45.py b/vllm/model_executor/models/ernie45.py index 2a89fffe80e35..e7302dc5ecdd7 100644 --- a/vllm/model_executor/models/ernie45.py +++ b/vllm/model_executor/models/ernie45.py @@ -28,7 +28,7 @@ from vllm.model_executor.models.llama import LlamaForCausalLM from .utils import PPMissingLayer -class Ernie4_5_ForCausalLM(LlamaForCausalLM): +class Ernie4_5ForCausalLM(LlamaForCausalLM): def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""): super().__init__(vllm_config=vllm_config, prefix=prefix) diff --git a/vllm/model_executor/models/registry.py b/vllm/model_executor/models/registry.py index 5e3a39a6cdd27..51831a770347a 100644 --- a/vllm/model_executor/models/registry.py +++ b/vllm/model_executor/models/registry.py @@ -60,7 +60,7 @@ _TEXT_GENERATION_MODELS = { "DeepseekV2ForCausalLM": ("deepseek_v2", "DeepseekV2ForCausalLM"), "DeepseekV3ForCausalLM": ("deepseek_v2", "DeepseekV3ForCausalLM"), "Dots1ForCausalLM": ("dots1", "Dots1ForCausalLM"), - "Ernie4_5_ForCausalLM": ("ernie45", "Ernie4_5_ForCausalLM"), + "Ernie4_5ForCausalLM": ("ernie45", "Ernie4_5ForCausalLM"), "Ernie4_5_MoeForCausalLM": ("ernie45_moe", "Ernie4_5_MoeForCausalLM"), "ExaoneForCausalLM": ("exaone", "ExaoneForCausalLM"), "Exaone4ForCausalLM": ("exaone4", "Exaone4ForCausalLM"),