diff --git a/docs/examples/llm/modelscope.ipynb b/docs/examples/llm/modelscope.ipynb
new file mode 100644
index 0000000000000000000000000000000000000000..8f19987a0802aca7399d1cdbe067e9ab30504cb8
--- /dev/null
+++ b/docs/examples/llm/modelscope.ipynb
@@ -0,0 +1,106 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "<a href=\"https://colab.research.google.com/github/run-llama/llama_index/blob/main/docs/examples/llm/modelscope.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# ModelScope LLMS"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "In this notebook, we show how to use the ModelScope LLM models in LlamaIndex. Check out the [ModelScope site](https://www.modelscope.cn/).\n",
+ "\n",
+ "If you're opening this Notebook on colab, you will need to install LlamaIndex 🦙 and the modelscope."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "!pip install llama-index-llms-modelscope"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Basic Usage\n"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import sys\n",
+ "from llama_index.llms.modelscope import ModelScopeLLM\n",
+ "\n",
+ "llm = ModelScopeLLM(model_name=\"qwen/Qwen1.5-7B-Chat\", model_revision=\"master\")\n",
+ "\n",
+ "rsp = llm.complete(\"Hello, who are you?\")\n",
+ "print(rsp)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "#### Use Message request"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "from llama_index.core.base.llms.types import MessageRole, ChatMessage\n",
+ "\n",
+ "messages = [\n",
+ " ChatMessage(\n",
+ " role=MessageRole.SYSTEM, content=\"You are a helpful assistant.\"\n",
+ " ),\n",
+ " ChatMessage(role=MessageRole.USER, content=\"How to make cake?\"),\n",
+ "]\n",
+ "resp = llm.chat(messages)\n",
+ "print(resp)"
+ ]
+ }
+ ],
+ "metadata": {
+ "colab": {
+ "name": "gemini.ipynb",
+ "toc_visible": true
+ },
+ "kernelspec": {
+ "display_name": "Python 3",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 0
+}
diff --git a/docs/module_guides/models/llms/modules.md b/docs/module_guides/models/llms/modules.md
index 07e93054f0a60c64ce198bb0e160cf840ab66723..9e1e0044bea050f479093cc2a686b5308eb8cd89 100644
--- a/docs/module_guides/models/llms/modules.md
+++ b/docs/module_guides/models/llms/modules.md
@@ -378,3 +378,12 @@ maxdepth: 1
---
/examples/llm/xinference_local_deployment.ipynb
```
+
+## ModelScope
+
+```{toctree}
+---
+maxdepth: 1
+---
+/examples/llm/modelscope.ipynb
+```
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/.gitignore b/llama-index-integrations/llms/llama-index-llms-modelscope/.gitignore
new file mode 100644
index 0000000000000000000000000000000000000000..990c18de229088f55c6c514fd0f2d49981d1b0e7
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/.gitignore
@@ -0,0 +1,153 @@
+llama_index/_static
+.DS_Store
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+bin/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+etc/
+include/
+lib/
+lib64/
+parts/
+sdist/
+share/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+# Usually these files are written by a python script from a template
+# before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+.ruff_cache
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+notebooks/
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+# According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+# However, in case of collaboration, if having platform-specific dependencies or dependencies
+# having no cross-platform support, pipenv may install dependencies that don't work, or not
+# install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+pyvenv.cfg
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# Jetbrains
+.idea
+modules/
+*.swp
+
+# VsCode
+.vscode
+
+# pipenv
+Pipfile
+Pipfile.lock
+
+# pyright
+pyrightconfig.json
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/BUILD b/llama-index-integrations/llms/llama-index-llms-modelscope/BUILD
new file mode 100644
index 0000000000000000000000000000000000000000..0896ca890d8bffd60a44fa824f8d57fecd73ee53
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/BUILD
@@ -0,0 +1,3 @@
+poetry_requirements(
+ name="poetry",
+)
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/Makefile b/llama-index-integrations/llms/llama-index-llms-modelscope/Makefile
new file mode 100644
index 0000000000000000000000000000000000000000..b9eab05aa370629a4a3de75df3ff64cd53887b68
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/Makefile
@@ -0,0 +1,17 @@
+GIT_ROOT ?= $(shell git rev-parse --show-toplevel)
+
+help: ## Show all Makefile targets.
+ @grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[33m%-30s\033[0m %s\n", $$1, $$2}'
+
+format: ## Run code autoformatters (black).
+ pre-commit install
+ git ls-files | xargs pre-commit run black --files
+
+lint: ## Run linters: pre-commit (black, ruff, codespell) and mypy
+ pre-commit install && git ls-files | xargs pre-commit run --show-diff-on-failure --files
+
+test: ## Run tests via pytest.
+ pytest tests
+
+watch-docs: ## Build and watch documentation.
+ sphinx-autobuild docs/ docs/_build/html --open-browser --watch $(GIT_ROOT)/llama_index/
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/README.md b/llama-index-integrations/llms/llama-index-llms-modelscope/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..58064ad24a95a26cb3a8a67192f68264d88bdb26
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/README.md
@@ -0,0 +1 @@
+# LlamaIndex Llms Integration: ModelScope
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/BUILD b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/BUILD
new file mode 100644
index 0000000000000000000000000000000000000000..db46e8d6c978c67e301dd6c47bee08c1b3fd141c
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/BUILD
@@ -0,0 +1 @@
+python_sources()
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/__init__.py b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..513713369f6957e15acde03ce9aca6e4012ef7b6
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/__init__.py
@@ -0,0 +1,3 @@
+from llama_index.llms.modelscope.base import ModelScopeLLM
+
+__all__ = ["ModelScopeLLM"]
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/base.py b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/base.py
new file mode 100644
index 0000000000000000000000000000000000000000..db26971e732a747727fc12c39655cbe30c7221b6
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/base.py
@@ -0,0 +1,180 @@
+import logging
+from typing import Any, Optional, Sequence
+import torch
+from llama_index.core.base.llms.types import (
+ ChatMessage,
+ ChatResponse,
+ ChatResponseGen,
+ CompletionResponse,
+ CompletionResponseGen,
+ LLMMetadata,
+)
+from llama_index.core.bridge.pydantic import Field, PrivateAttr
+from llama_index.core.callbacks import CallbackManager
+from llama_index.core.constants import (
+ DEFAULT_CONTEXT_WINDOW,
+ DEFAULT_NUM_OUTPUTS,
+)
+from llama_index.core.llms.callbacks import (
+ llm_chat_callback,
+ llm_completion_callback,
+)
+from llama_index.core.llms.custom import CustomLLM
+from llama_index.core.prompts.base import PromptTemplate
+from llama_index.core.types import PydanticProgramMode
+from llama_index.llms.modelscope.utils import (
+ chat_message_to_modelscope_messages,
+ text_to_completion_response,
+ modelscope_message_to_chat_response,
+)
+from modelscope import pipeline
+
+DEFAULT_MODELSCOPE_MODEL = "qwen/Qwen-7B-Chat"
+DEFAULT_MODELSCOPE_MODEL_REVISION = "master"
+DEFAULT_MODELSCOPE_TASK = "chat"
+DEFAULT_MODELSCOPE_DTYPE = "float16"
+logger = logging.getLogger(__name__)
+
+_STR_DTYPE_TO_TORCH_DTYPE = {
+ "half": torch.float16,
+ "float16": torch.float16,
+ "float": torch.float32,
+ "float32": torch.float32,
+ "bfloat16": torch.bfloat16,
+}
+
+
+class ModelScopeLLM(CustomLLM):
+ """ModelScope LLM."""
+
+ model_name: str = Field(
+ default=DEFAULT_MODELSCOPE_MODEL,
+ description=(
+ "The model name to use from ModelScope. "
+ "Unused if `model` is passed in directly."
+ ),
+ )
+ model_revision: str = Field(
+ default=DEFAULT_MODELSCOPE_MODEL_REVISION,
+ description=(
+ "The model revision to use from ModelScope. "
+ "Unused if `model` is passed in directly."
+ ),
+ )
+ task_name: str = Field(
+ default=DEFAULT_MODELSCOPE_TASK,
+ description=("The ModelScope task type, for llm use default chat."),
+ )
+ dtype: str = Field(
+ default=DEFAULT_MODELSCOPE_DTYPE,
+ description=("The ModelScope task type, for llm use default chat."),
+ )
+ context_window: int = Field(
+ default=DEFAULT_CONTEXT_WINDOW,
+ description="The maximum number of tokens available for input.",
+ gt=0,
+ )
+ max_new_tokens: int = Field(
+ default=DEFAULT_NUM_OUTPUTS,
+ description="The maximum number of tokens to generate.",
+ gt=0,
+ )
+ system_prompt: str = Field(
+ default="",
+ description=(
+ "The system prompt, containing any extra instructions or context. "
+ "The model card on ModelScope should specify if this is needed."
+ ),
+ )
+ query_wrapper_prompt: PromptTemplate = Field(
+ default=PromptTemplate("{query_str}"),
+ description=(
+ "The query wrapper prompt, containing the query placeholder. "
+ "The model card on ModelScope should specify if this is needed. "
+ "Should contain a `{query_str}` placeholder."
+ ),
+ )
+ device_map: str = Field(
+ default="auto", description="The device_map to use. Defaults to 'auto'."
+ )
+ tokenizer_kwargs: dict = Field(
+ default_factory=dict, description="The kwargs to pass to the tokenizer."
+ )
+ model_kwargs: dict = Field(
+ default_factory=dict,
+ description="The kwargs to pass to the model during initialization.",
+ )
+ generate_kwargs: dict = Field(
+ default_factory=dict,
+ description="The kwargs to pass to the model during generation.",
+ )
+
+ _pipeline: Any = PrivateAttr()
+
+ def __init__(
+ self,
+ model_name: str = DEFAULT_MODELSCOPE_MODEL,
+ model_revision: str = DEFAULT_MODELSCOPE_MODEL_REVISION,
+ task_name: str = DEFAULT_MODELSCOPE_TASK,
+ dtype: str = DEFAULT_MODELSCOPE_DTYPE,
+ model: Optional[Any] = None,
+ device_map: Optional[str] = "auto",
+ model_kwargs: Optional[dict] = None,
+ generate_kwargs: Optional[dict] = None,
+ callback_manager: Optional[CallbackManager] = None,
+ pydantic_program_mode: PydanticProgramMode = PydanticProgramMode.DEFAULT,
+ ) -> None:
+ """Initialize params."""
+ model_kwargs = model_kwargs or {}
+ if model:
+ self._pipeline = model
+ else:
+ self._pipeline = pipeline(
+ task=task_name,
+ model=model_name,
+ model_revision=model_revision,
+ llm_first=True,
+ torch_dtype=_STR_DTYPE_TO_TORCH_DTYPE[dtype],
+ device_map=device_map,
+ )
+
+ super().__init__(
+ model_kwargs=model_kwargs or {},
+ generate_kwargs=generate_kwargs or {},
+ callback_manager=callback_manager,
+ pydantic_program_mode=pydantic_program_mode,
+ )
+
+ @classmethod
+ def class_name(cls) -> str:
+ return "ModelScope_LLM"
+
+ @property
+ def metadata(self) -> LLMMetadata:
+ """LLM metadata."""
+ return LLMMetadata(
+ context_window=None,
+ num_output=None,
+ model_name=self.model_name,
+ is_chat_model=self.is_chat_model,
+ )
+
+ @llm_completion_callback()
+ def complete(self, prompt: str, **kwargs: Any) -> CompletionResponse:
+ return text_to_completion_response(self._pipeline(prompt, **kwargs))
+
+ @llm_completion_callback()
+ def stream_complete(self, prompt: str, **kwargs: Any) -> CompletionResponseGen:
+ yield self.complete(prompt, **kwargs)
+
+ @llm_chat_callback()
+ def chat(self, messages: Sequence[ChatMessage], **kwargs: Any) -> ChatResponse:
+ return modelscope_message_to_chat_response(
+ self._pipeline(chat_message_to_modelscope_messages(messages), **kwargs)
+ )
+
+ @llm_chat_callback()
+ def stream_chat(
+ self, messages: Sequence[ChatMessage], **kwargs: Any
+ ) -> ChatResponseGen:
+ yield self.chat(messages, **kwargs)
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/utils.py b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..2152822db464da3d9caf2edd16f1902ee0a95491
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/llama_index/llms/modelscope/utils.py
@@ -0,0 +1,30 @@
+from typing import Dict, List, Sequence
+
+from llama_index.core.base.llms.types import (
+ ChatResponse,
+ CompletionResponse,
+ ChatMessage,
+)
+
+
+def chat_message_to_modelscope_messages(
+ chat_messages: Sequence[ChatMessage],
+) -> List[Dict]:
+ messages = []
+ for msg in chat_messages:
+ messages.append({"role": msg.role.value, "content": msg.content})
+ return {"messages": messages}
+
+
+def text_to_completion_response(output) -> CompletionResponse:
+ return CompletionResponse(text=output["text"], raw=output)
+
+
+def modelscope_message_to_chat_response(output) -> ChatResponse:
+ # output format: {'message': {'role': 'assistant', 'content': ''}}
+ return ChatResponse(
+ message=ChatMessage(
+ role=output["message"]["role"], content=output["message"]["content"]
+ ),
+ raw=output,
+ )
diff --git a/llama-index-integrations/llms/llama-index-llms-modelscope/pyproject.toml b/llama-index-integrations/llms/llama-index-llms-modelscope/pyproject.toml
new file mode 100644
index 0000000000000000000000000000000000000000..248cb8eeb4b1c06d0387305e7c58ba8b351b30e6
--- /dev/null
+++ b/llama-index-integrations/llms/llama-index-llms-modelscope/pyproject.toml
@@ -0,0 +1,68 @@
+[build-system]
+build-backend = "poetry.core.masonry.api"
+requires = ["poetry-core"]
+
+[tool.codespell]
+check-filenames = true
+check-hidden = true
+skip = "*.csv,*.html,*.json,*.jsonl,*.pdf,*.txt,*.ipynb"
+
+[tool.llamahub]
+contains_example = false
+import_path = "llama_index.llms.modelscope"
+
+[tool.llamahub.class_authors]
+ModelScopeLLM = "llama-index"
+
+[tool.mypy]
+disallow_untyped_defs = true
+exclude = ["_static", "build", "examples", "notebooks", "venv"]
+ignore_missing_imports = true
+python_version = "3.8"
+
+[tool.poetry]
+authors = ["ModelScope <modelscope@list.alibaba-inc.com>"]
+description = "llama-index llms modelscope integration"
+exclude = ["**/BUILD"]
+license = "MIT"
+name = "llama-index-llms-modelscope"
+readme = "README.md"
+version = "0.1.1"
+
+[tool.poetry.dependencies]
+python = ">=3.8.1,<3.12"
+llama-index-core = "^0.10.1"
+modelscope = ">=1.12.0"
+torch = "^2.1.2"
+
+[tool.poetry.dependencies.transformers]
+extras = ["torch"]
+version = "^4.37.0"
+
+[tool.poetry.group.dev.dependencies]
+ipython = "8.10.0"
+jupyter = "^1.0.0"
+mypy = "0.991"
+pre-commit = "3.2.0"
+pylint = "2.15.10"
+pytest = "7.2.1"
+pytest-mock = "3.11.1"
+ruff = "0.0.292"
+tree-sitter-languages = "^1.8.0"
+types-Deprecated = ">=0.1.0"
+types-PyYAML = "^6.0.12.12"
+types-protobuf = "^4.24.0.4"
+types-redis = "4.5.5.0"
+types-requests = "2.28.11.8"
+types-setuptools = "67.1.0.0"
+
+[tool.poetry.group.dev.dependencies.black]
+extras = ["jupyter"]
+version = "<=23.9.1,>=23.7.0"
+
+[tool.poetry.group.dev.dependencies.codespell]
+extras = ["toml"]
+version = ">=v2.2.6"
+
+[[tool.poetry.packages]]
+include = "llama_index/"