Skip to content

[template] support mimo-v2 template #7095

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Jintao-Huang wants to merge 6 commits into
base: main
Choose a base branch
from
Open

[template] support mimo-v2 template #7095

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
7969b9e
support mimo v2 template
Jintao-Huang Dec 17, 2025
db4b8ba
fix
Jintao-Huang Dec 17, 2025
24bcd34
Merge branch 'main' into support_mimo_v2_template
Jintao-Huang Dec 18, 2025
7946db2
Merge branch 'main' into support_mimo_v2_template
Jintao-Huang Dec 19, 2025
7172f86
Merge branch 'main' into support_mimo_v2_template
Jintao-Huang Dec 21, 2025
e1a0780
Merge remote-tracking branch 'refs/remotes/origin/support_mimo_v2_tem…
Jintao-Huang Dec 21, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions docs/source/Instruction/Supported-models-and-datasets.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -647,6 +647,8 @@
|[PaddlePaddle/ERNIE-4.5-21B-A3B-Thinking](https://modelscope.cn/models/PaddlePaddle/ERNIE-4.5-21B-A3B-Thinking)|ernie_thinking|ernie_thinking|-|✔|-|[baidu/ERNIE-4.5-21B-A3B-Thinking](https://huggingface.co/baidu/ERNIE-4.5-21B-A3B-Thinking)|
|[meituan-longcat/LongCat-Flash-Chat](https://modelscope.cn/models/meituan-longcat/LongCat-Flash-Chat)|longchat|longchat|transformers>=4.54,<4.56|&#x2718;|-|[meituan-longcat/LongCat-Flash-Chat](https://huggingface.co/meituan-longcat/LongCat-Flash-Chat)|
|[meituan-longcat/LongCat-Flash-Chat-FP8](https://modelscope.cn/models/meituan-longcat/LongCat-Flash-Chat-FP8)|longchat|longchat|transformers>=4.54,<4.56|&#x2718;|-|[meituan-longcat/LongCat-Flash-Chat-FP8](https://huggingface.co/meituan-longcat/LongCat-Flash-Chat-FP8)|
|[XiaomiMiMo/MiMo-V2-Flash](https://modelscope.cn/models/XiaomiMiMo/MiMo-V2-Flash)|mimo_v2|mimo_v2|-|&#x2718;|-|[XiaomiMiMo/MiMo-V2-Flash](https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash)|
|[XiaomiMiMo/MiMo-V2-Flash-Base](https://modelscope.cn/models/XiaomiMiMo/MiMo-V2-Flash-Base)|mimo_v2|mimo_v2|-|&#x2718;|-|[XiaomiMiMo/MiMo-V2-Flash-Base](https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash-Base)|
|[answerdotai/ModernBERT-base](https://modelscope.cn/models/answerdotai/ModernBERT-base)|modern_bert|dummy|transformers>=4.48|&#x2718;|bert|[answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base)|
|[answerdotai/ModernBERT-large](https://modelscope.cn/models/answerdotai/ModernBERT-large)|modern_bert|dummy|transformers>=4.48|&#x2718;|bert|[answerdotai/ModernBERT-large](https://huggingface.co/answerdotai/ModernBERT-large)|
|[iic/gte-modernbert-base](https://modelscope.cn/models/iic/gte-modernbert-base)|modern_bert_gte|dummy|transformers>=4.48|&#x2718;|bert, embedding|[Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base)|
Expand Down
2 changes: 2 additions & 0 deletions docs/source_en/Instruction/Supported-models-and-datasets.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -649,6 +649,8 @@ The table below introduces the models integrated with ms-swift:
|[PaddlePaddle/ERNIE-4.5-21B-A3B-Thinking](https://modelscope.cn/models/PaddlePaddle/ERNIE-4.5-21B-A3B-Thinking)|ernie_thinking|ernie_thinking|-|&#x2714;|-|[baidu/ERNIE-4.5-21B-A3B-Thinking](https://huggingface.co/baidu/ERNIE-4.5-21B-A3B-Thinking)|
|[meituan-longcat/LongCat-Flash-Chat](https://modelscope.cn/models/meituan-longcat/LongCat-Flash-Chat)|longchat|longchat|transformers>=4.54,<4.56|&#x2718;|-|[meituan-longcat/LongCat-Flash-Chat](https://huggingface.co/meituan-longcat/LongCat-Flash-Chat)|
|[meituan-longcat/LongCat-Flash-Chat-FP8](https://modelscope.cn/models/meituan-longcat/LongCat-Flash-Chat-FP8)|longchat|longchat|transformers>=4.54,<4.56|&#x2718;|-|[meituan-longcat/LongCat-Flash-Chat-FP8](https://huggingface.co/meituan-longcat/LongCat-Flash-Chat-FP8)|
|[XiaomiMiMo/MiMo-V2-Flash](https://modelscope.cn/models/XiaomiMiMo/MiMo-V2-Flash)|mimo_v2|mimo_v2|-|&#x2718;|-|[XiaomiMiMo/MiMo-V2-Flash](https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash)|
|[XiaomiMiMo/MiMo-V2-Flash-Base](https://modelscope.cn/models/XiaomiMiMo/MiMo-V2-Flash-Base)|mimo_v2|mimo_v2|-|&#x2718;|-|[XiaomiMiMo/MiMo-V2-Flash-Base](https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash-Base)|
|[answerdotai/ModernBERT-base](https://modelscope.cn/models/answerdotai/ModernBERT-base)|modern_bert|dummy|transformers>=4.48|&#x2718;|bert|[answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base)|
|[answerdotai/ModernBERT-large](https://modelscope.cn/models/answerdotai/ModernBERT-large)|modern_bert|dummy|transformers>=4.48|&#x2718;|bert|[answerdotai/ModernBERT-large](https://huggingface.co/answerdotai/ModernBERT-large)|
|[iic/gte-modernbert-base](https://modelscope.cn/models/iic/gte-modernbert-base)|modern_bert_gte|dummy|transformers>=4.48|&#x2718;|bert, embedding|[Alibaba-NLP/gte-modernbert-base](https://huggingface.co/Alibaba-NLP/gte-modernbert-base)|
Expand Down
1 change: 1 addition & 0 deletions swift/llm/model/constant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,7 @@ class LLMModelType:
gemma_emb = 'gemma_emb'
ernie_thinking = 'ernie_thinking'
longchat = 'longchat'
mimo_v2 = 'mimo_v2'


class BertModelType:
Expand Down
14 changes: 14 additions & 0 deletions swift/llm/model/model/llm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -397,3 +397,17 @@ def get_model_tokenizer_yuan(model_dir: str,
get_model_tokenizer_with_flash_attn,
architectures=['BailingMoeV2ForCausalLM'],
))

register_model(
ModelMeta(
LLMModelType.mimo_v2,
[
ModelGroup([
Model('XiaomiMiMo/MiMo-V2-Flash', 'XiaomiMiMo/MiMo-V2-Flash'),
Model('XiaomiMiMo/MiMo-V2-Flash-Base', 'XiaomiMiMo/MiMo-V2-Flash-Base'),
])
],
TemplateType.mimo_v2,
get_model_tokenizer_with_flash_attn,
architectures=['MiMoV2FlashForCausalLM'],
))
1 change: 1 addition & 0 deletions swift/llm/template/constant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,6 +104,7 @@ class LLMTemplateType:
ernie = 'ernie'
ernie_thinking = 'ernie_thinking'
longchat = 'longchat'
mimo_v2 = 'mimo_v2'

aya = 'aya'
c4ai = 'c4ai'
Expand Down
7 changes: 7 additions & 0 deletions swift/llm/template/template/llm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -424,3 +424,10 @@ class GptOssTemplateMeta(TemplateMeta):
is_thinking=True,
thinking_prefix='<think>\n',
))

register_template(
ChatmlTemplateMeta(
LLMTemplateType.mimo_v2,
default_system='You are MiMo, a helpful AI assistant engineered by Xiaomi.',
response_prefix='<think></think>',
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Dec 17, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The response_prefix is set to <think></think>, which is an empty tag pair. This is inconsistent with other "thinking" templates like ring2 and gpt_oss which use <think>\n to allow the model to generate thoughts. An empty tag might not be the intended behavior. If the model is not expected to generate thoughts, it might be better to use <think>\n</think>\n to represent an empty thinking step explicitly. If it is expected to generate thoughts, <think>\n would be more appropriate.

Suggested change
response_prefix='<think></think>',
response_prefix='<think>\n</think>\n',

))
Loading