frozenleaves/vllm-dev
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[Misc] Modify the organization of GLM series (#22171)

Browse Source 
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
This commit is contained in:
Jee Jee Li
2025-08-04 14:51:20 +08:00
committed by GitHub
parent 8ecb3e9e93
commit a7b8788d2c
16 changed files with 31 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
docs/models/supported_models.md
View File

@ -328,7 +328,7 @@ th {
| `BambaForCausalLM` | Bamba | `ibm-ai-platform/Bamba-9B-fp8`, `ibm-ai-platform/Bamba-9B` | ✅︎ | ✅︎ | ✅︎ |
| `BloomForCausalLM` | BLOOM, BLOOMZ, BLOOMChat | `bigscience/bloom`, `bigscience/bloomz`, etc. | | ✅︎ | |
| `BartForConditionalGeneration` | BART | `facebook/bart-base`, `facebook/bart-large-cnn`, etc. | | | |
| `ChatGLMModel`, `ChatGLMForConditionalGeneration` | ChatGLM | `THUDM/chatglm2-6b`, `THUDM/chatglm3-6b`, `ShieldLM-6B-chatglm3`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `ChatGLMModel`, `ChatGLMForConditionalGeneration` | ChatGLM | `zai-org/chatglm2-6b`, `zai-org/chatglm3-6b`, `ShieldLM-6B-chatglm3`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `CohereForCausalLM`, `Cohere2ForCausalLM` | Command-R | `CohereForAI/c4ai-command-r-v01`, `CohereForAI/c4ai-command-r7b-12-2024`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `DbrxForCausalLM` | DBRX | `databricks/dbrx-base`, `databricks/dbrx-instruct`, etc. | | ✅︎ | ✅︎ |
| `DeciLMForCausalLM` | DeciLM | `nvidia/Llama-3_3-Nemotron-Super-49B-v1`, etc. | ✅︎ | ✅︎ | ✅︎ |
@ -348,8 +348,8 @@ th {
| `Gemma2ForCausalLM` | Gemma 2 | `google/gemma-2-9b`, `google/gemma-2-27b`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Gemma3ForCausalLM` | Gemma 3 | `google/gemma-3-1b-it`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Gemma3nForConditionalGeneration` | Gemma 3n | `google/gemma-3n-E2B-it`, `google/gemma-3n-E4B-it`, etc. | | | ✅︎ |
| `GlmForCausalLM` | GLM-4 | `THUDM/glm-4-9b-chat-hf`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4ForCausalLM` | GLM-4-0414 | `THUDM/GLM-4-32B-0414`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GlmForCausalLM` | GLM-4 | `zai-org/glm-4-9b-chat-hf`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4ForCausalLM` | GLM-4-0414 | `zai-org/GLM-4-32B-0414`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GPT2LMHeadModel` | GPT-2 | `gpt2`, `gpt2-xl`, etc. | | ✅︎ | ✅︎ |
| `GPTBigCodeForCausalLM` | StarCoder, SantaCoder, WizardCoder | `bigcode/starcoder`, `bigcode/gpt_bigcode-santacoder`, `WizardLM/WizardCoder-15B-V1.0`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GPTJForCausalLM` | GPT-J | `EleutherAI/gpt-j-6b`, `nomic-ai/gpt4all-j`, etc. | | ✅︎ | ✅︎ |
@ -589,8 +589,8 @@ See [this page](generative_models.md) for more information on how to use generat
| `Florence2ForConditionalGeneration` | Florence-2 | T + I | `microsoft/Florence-2-base`, `microsoft/Florence-2-large`, etc. | | | |
| `FuyuForCausalLM` | Fuyu | T + I | `adept/fuyu-8b`, etc. | | ✅︎ | ✅︎ |
| `Gemma3ForConditionalGeneration` | Gemma 3 | T + I<sup>+</sup> | `google/gemma-3-4b-it`, `google/gemma-3-27b-it`, etc. | ✅︎ | ✅︎ | ⚠️ |
| `GLM4VForCausalLM`<sup>^</sup> | GLM-4V | T + I | `THUDM/glm-4v-9b`, `THUDM/cogagent-9b-20241220`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4vForConditionalGeneration` | GLM-4.1V-Thinking | T + I<sup>E+</sup> + V<sup>E+</sup> | `THUDM/GLM-4.1V-9B-Thinking`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GLM4VForCausalLM`<sup>^</sup> | GLM-4V | T + I | `zai-org/glm-4v-9b`, `zai-org/cogagent-9b-20241220`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4vForConditionalGeneration` | GLM-4.1V-Thinking | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.1V-9B-Thinking`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4MoeForCausalLM` | GLM-4.5 | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Glm4v_moeForConditionalGeneration` | GLM-4.5V | T + I<sup>E+</sup> + V<sup>E+</sup> | `zai-org/GLM-4.5V-Air`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `GraniteSpeechForConditionalGeneration` | Granite Speech | T + A | `ibm-granite/granite-speech-3.3-8b` | ✅︎ | ✅︎ | ✅︎ |

4
examples/offline_inference/vision_language.py
View File

@ -221,7 +221,7 @@ def run_gemma3(questions: list[str], modality: str) -> ModelRequestData:
# GLM-4v
def run_glm4v(questions: list[str], modality: str) -> ModelRequestData:
assert modality == "image"
model_name = "THUDM/glm-4v-9b"
model_name = "zai-org/glm-4v-9b"
engine_args = EngineArgs(
model=model_name,
@ -250,7 +250,7 @@ def run_glm4v(questions: list[str], modality: str) -> ModelRequestData:
# GLM-4.1V
def run_glm4_1v(questions: list[str], modality: str) -> ModelRequestData:
model_name = "THUDM/GLM-4.1V-9B-Thinking"
model_name = "zai-org/GLM-4.1V-9B-Thinking"
engine_args = EngineArgs(
model=model_name,

4
tests/distributed/test_pipeline_parallel.py
View File

@ -154,7 +154,7 @@ TEXT_GENERATION_MODELS = {
"baichuan-inc/Baichuan-7B": PPTestSettings.fast(),
"baichuan-inc/Baichuan2-13B-Chat": PPTestSettings.fast(),
"bigscience/bloomz-1b1": PPTestSettings.fast(),
"THUDM/chatglm3-6b": PPTestSettings.fast(),
"zai-org/chatglm3-6b": PPTestSettings.fast(),
"CohereForAI/c4ai-command-r-v01": PPTestSettings.fast(load_format="dummy"),
"databricks/dbrx-instruct": PPTestSettings.fast(load_format="dummy"),
"Deci/DeciLM-7B-instruct": PPTestSettings.fast(),
@ -224,7 +224,7 @@ MULTIMODAL_MODELS = {
"Salesforce/blip2-opt-6.7b": PPTestSettings.fast(),
"facebook/chameleon-7b": PPTestSettings.fast(),
"adept/fuyu-8b": PPTestSettings.fast(),
"THUDM/glm-4v-9b": PPTestSettings.fast(),
"zai-org/glm-4v-9b": PPTestSettings.fast(),
"OpenGVLab/InternVL2-1B": PPTestSettings.fast(),
"llava-hf/llava-1.5-7b-hf": PPTestSettings.fast(),
"llava-hf/llava-v1.6-mistral-7b-hf": PPTestSettings.fast(),

2
tests/lora/test_add_lora.py
View File

@ -14,7 +14,7 @@ from vllm.lora.request import LoRARequest
from vllm.sampling_params import SamplingParams
from vllm.utils import merge_async_iterators
MODEL_PATH = "THUDM/chatglm3-6b"
MODEL_PATH = "zai-org/chatglm3-6b"
LORA_RANK = 64
DEFAULT_MAX_LORAS = 4 * 3

2
tests/lora/test_chatglm3_tp.py
View File

@ -6,7 +6,7 @@ from vllm.lora.request import LoRARequest
from ..utils import create_new_process_for_each_test, multi_gpu_test
MODEL_PATH = "THUDM/chatglm3-6b"
MODEL_PATH = "zai-org/chatglm3-6b"
PROMPT_TEMPLATE = """I want you to act as a SQL terminal in front of an example database, you need only to return the sql command to me.Below is an instruction that describes a task, Write a response that appropriately completes the request.\n"\n##Instruction:\nconcert_singer contains tables such as stadium, singer, concert, singer_in_concert. Table stadium has columns such as Stadium_ID, Location, Name, Capacity, Highest, Lowest, Average. Stadium_ID is the primary key.\nTable singer has columns such as Singer_ID, Name, Country, Song_Name, Song_release_year, Age, Is_male. Singer_ID is the primary key.\nTable concert has columns such as concert_ID, concert_Name, Theme, Stadium_ID, Year. concert_ID is the primary key.\nTable singer_in_concert has columns such as concert_ID, Singer_ID. concert_ID is the primary key.\nThe Stadium_ID of concert is the foreign key of Stadium_ID of stadium.\nThe Singer_ID of singer_in_concert is the foreign key of Singer_ID of singer.\nThe concert_ID of singer_in_concert is the foreign key of concert_ID of concert.\n\n###Input:\n{query}\n\n###Response:""" # noqa: E501

2
tests/models/language/generation/test_common.py
View File

@ -53,7 +53,7 @@ AITER_MODEL_LIST = [
marks=[pytest.mark.core_model, pytest.mark.cpu_model],
),
pytest.param(
"THUDM/chatglm3-6b", # chatglm (text-only)
"zai-org/chatglm3-6b", # chatglm (text-only)
),
pytest.param(
"meta-llama/Llama-3.2-1B-Instruct", # llama

6
tests/models/multimodal/generation/test_common.py
View File

@ -355,7 +355,7 @@ VLM_TEST_SETTINGS = {
num_logprobs=10,
),
"glm4v": VLMTestInfo(
models=["THUDM/glm-4v-9b"],
models=["zai-org/glm-4v-9b"],
test_type=VLMTestType.IMAGE,
prompt_formatter=lambda img_prompt: f"<|user|>\n{img_prompt}<|assistant|>", # noqa: E501
single_image_prompts=IMAGE_ASSETS.prompts({
@ -374,7 +374,7 @@ VLM_TEST_SETTINGS = {
marks=[large_gpu_mark(min_gb=32)],
),
"glm4_1v": VLMTestInfo(
models=["THUDM/GLM-4.1V-9B-Thinking"],
models=["zai-org/GLM-4.1V-9B-Thinking"],
test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),
prompt_formatter=lambda img_prompt: f"<|user|>\n{img_prompt}<|assistant|>", # noqa: E501
img_idx_to_prompt=lambda idx: "<|begin_of_image|><|image|><|end_of_image|>", # noqa: E501
@ -388,7 +388,7 @@ VLM_TEST_SETTINGS = {
marks=[large_gpu_mark(min_gb=32)],
),
"glm4_1v-video": VLMTestInfo(
models=["THUDM/GLM-4.1V-9B-Thinking"],
models=["zai-org/GLM-4.1V-9B-Thinking"],
# GLM4.1V require include video metadata for input
test_type=VLMTestType.CUSTOM_INPUTS,
max_model_len=4096,

4
tests/models/multimodal/processing/test_common.py
View File

@ -271,8 +271,8 @@ def _test_processing_correctness_one(
"microsoft/Florence-2-base",
"adept/fuyu-8b",
"google/gemma-3-4b-it",
"THUDM/glm-4v-9b",
"THUDM/GLM-4.1V-9B-Thinking",
"zai-org/glm-4v-9b",
"zai-org/GLM-4.1V-9B-Thinking",
"ibm-granite/granite-speech-3.3-2b",
"h2oai/h2ovl-mississippi-800m",
"internlm/Intern-S1",

2
tests/models/multimodal/processing/test_glm4_1v.py
View File

@ -9,7 +9,7 @@ from vllm.multimodal import MULTIMODAL_REGISTRY
from ...utils import build_model_context
@pytest.mark.parametrize("model_id", ["THUDM/GLM-4.1V-9B-Thinking"])
@pytest.mark.parametrize("model_id", ["zai-org/GLM-4.1V-9B-Thinking"])
@pytest.mark.parametrize("expected_toks_per_frame", [299])
@pytest.mark.parametrize("num_frames", [32, 128])
@pytest.mark.parametrize("fps, expected_grid_t", [(1, 5), (2, 10)])

10
tests/models/registry.py
View File

@ -153,7 +153,7 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
extras={"tiny": "hmellor/tiny-random-BambaForCausalLM"}), # noqa: E501
"BloomForCausalLM": _HfExamplesInfo("bigscience/bloom-560m",
{"1b": "bigscience/bloomz-1b1"}),
"ChatGLMModel": _HfExamplesInfo("THUDM/chatglm3-6b",
"ChatGLMModel": _HfExamplesInfo("zai-org/chatglm3-6b",
trust_remote_code=True,
max_transformers_version="4.48"),
"ChatGLMForConditionalGeneration": _HfExamplesInfo("thu-coai/ShieldLM-6B-chatglm3", # noqa: E501
@ -187,8 +187,8 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
"Gemma3ForCausalLM": _HfExamplesInfo("google/gemma-3-1b-it"),
"Gemma3nForConditionalGeneration": _HfExamplesInfo("google/gemma-3n-E2B-it", # noqa: E501
min_transformers_version="4.53"),
"GlmForCausalLM": _HfExamplesInfo("THUDM/glm-4-9b-chat-hf"),
"Glm4ForCausalLM": _HfExamplesInfo("THUDM/GLM-4-9B-0414"),
"GlmForCausalLM": _HfExamplesInfo("zai-org/glm-4-9b-chat-hf"),
"Glm4ForCausalLM": _HfExamplesInfo("zai-org/GLM-4-9B-0414"),
"Glm4MoeForCausalLM": _HfExamplesInfo("zai-org/GLM-4.5",
min_transformers_version="4.54"), # noqa: E501
"GPT2LMHeadModel": _HfExamplesInfo("openai-community/gpt2",
@ -380,10 +380,10 @@ _MULTIMODAL_EXAMPLE_MODELS = {
"FuyuForCausalLM": _HfExamplesInfo("adept/fuyu-8b"),
"Gemma3ForConditionalGeneration": _HfExamplesInfo("google/gemma-3-4b-it"),
"GraniteSpeechForConditionalGeneration": _HfExamplesInfo("ibm-granite/granite-speech-3.3-2b"), # noqa: E501
"GLM4VForCausalLM": _HfExamplesInfo("THUDM/glm-4v-9b",
"GLM4VForCausalLM": _HfExamplesInfo("zai-org/glm-4v-9b",
trust_remote_code=True,
hf_overrides={"architectures": ["GLM4VForCausalLM"]}), # noqa: E501
"Glm4vForConditionalGeneration": _HfExamplesInfo("THUDM/GLM-4.1V-9B-Thinking"), # noqa: E501
"Glm4vForConditionalGeneration": _HfExamplesInfo("zai-org/GLM-4.1V-9B-Thinking"), # noqa: E501
"Glm4v_moeForConditionalGeneration": _HfExamplesInfo("zai-org/GLM-4.5V-Air",
is_available_online=False), # noqa: E501
"H2OVLChatModel": _HfExamplesInfo("h2oai/h2ovl-mississippi-800m",

2
tests/tokenization/test_cached_tokenizer.py
View File

@ -10,7 +10,7 @@ from vllm.transformers_utils.tokenizer import (AnyTokenizer,
get_cached_tokenizer)
@pytest.mark.parametrize("model_id", ["gpt2", "THUDM/chatglm3-6b"])
@pytest.mark.parametrize("model_id", ["gpt2", "zai-org/chatglm3-6b"])
def test_cached_tokenizer(model_id: str):
reference_tokenizer = AutoTokenizer.from_pretrained(model_id,
trust_remote_code=True)

6
vllm/model_executor/models/chatglm.py
View File

@ -1,7 +1,7 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# Adapted from
# https://github.com/THUDM/ChatGLM2-6B
# https://github.com/zai-org/ChatGLM2-6B
"""Inference-only ChatGLM model compatible with THUDM weights."""
import json
from collections.abc import Iterable
@ -86,10 +86,10 @@ class GLMAttention(nn.Module):
prefix=f"{prefix}.dense",
)
# https://huggingface.co/THUDM/chatglm3-6b-32k/blob/e210410255278dd9d74463cf396ba559c0ef801c/modeling_chatglm.py#L141
# https://huggingface.co/zai-org/chatglm3-6b-32k/blob/e210410255278dd9d74463cf396ba559c0ef801c/modeling_chatglm.py#L141
rope_ratio = getattr(config, "rope_ratio", 1.0)
max_positions = getattr(config, "seq_length", 8192)
# NOTE: THUDM/cogagent-9b-20241220 uses original_rope=False,
# NOTE: zai-org/cogagent-9b-20241220 uses original_rope=False,
# which is equivalent to is_neox_style=True
is_neox_style = not config.original_rope
self.rotary_emb = get_rope(

2
vllm/model_executor/models/glm4v.py
View File

@ -2,7 +2,7 @@
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# Adapted from
# https://github.com/THUDM/CogAgent
# https://github.com/zai-org/CogAgent
"""Inference-only CogAgent model compatible with THUDM weights."""
from argparse import Namespace
from collections.abc import Mapping, Sequence

2
vllm/test_utils.py
View File

@ -118,7 +118,7 @@ MODELS_ON_S3 = [
"stabilityai/stablelm-zephyr-3b",
"state-spaces/mamba-130m-hf",
"TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ",
"THUDM/glm-4v-9b",
"zai-org/glm-4v-9b",
"TIGER-Lab/Mantis-8B-siglip-llama3",
"TIGER-Lab/VLM2Vec-Full",
"tiiuae/falcon-40b",

2
vllm/transformers_utils/configs/chatglm.py
View File

@ -2,7 +2,7 @@
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# Adapted from
# https://github.com/THUDM/ChatGLM2-6B
# https://github.com/zai-org/ChatGLM2-6B
from transformers import PretrainedConfig

2
vllm/transformers_utils/tokenizer.py
View File

@ -271,7 +271,7 @@ def get_tokenizer(
}
tokenizer.add_special_tokens(special_tokens_map)
# NOTE: We can remove this after https://github.com/THUDM/ChatGLM3/issues/1324
# NOTE: We can remove this after https://github.com/zai-org/ChatGLM3/issues/1324
if type(tokenizer).__name__ in ("ChatGLMTokenizer",
"ChatGLM4Tokenizer"):
assert isinstance(tokenizer, PreTrainedTokenizer)