From e2320e517b8f3e3edb76263c0576ce651f20adde Mon Sep 17 00:00:00 2001
From: MING_X <119648793+MING-ZCH@users.noreply.github.com>
Date: Fri, 27 Sep 2024 12:13:34 +0800
Subject: [PATCH 1/5] [docs] fix some bugs in docs.md (#249)

* Update quickstart.md

* Update overview.md

* Update overview.md
---
 docs/en/get_started/overview.md      | 2 +-
 docs/zh_cn/get_started/overview.md   | 2 +-
 docs/zh_cn/get_started/quickstart.md | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/docs/en/get_started/overview.md b/docs/en/get_started/overview.md
index 356b1481..5be46835 100644
--- a/docs/en/get_started/overview.md
+++ b/docs/en/get_started/overview.md
@@ -11,7 +11,7 @@ Lagent is an open source LLM agent framework, which enables people to efficientl
 Lagent consists of 3 main parts, agents, llms, and actions.
 
 - **agents** provides agent implementation, such as ReAct, AutoGPT.
-- **llms** supports various large language models, including open-sourced models (Llama-2, InterLM) through HuggingFace models or closed-source models like GPT3.5/4.
+- **llms** supports various large language models, including open-sourced models (Llama-2, InternLM) through HuggingFace models or closed-source models like GPT3.5/4.
 - **actions** contains a series of actions, as well as an action executor to manage all actions.
 
 ## How to Use
diff --git a/docs/zh_cn/get_started/overview.md b/docs/zh_cn/get_started/overview.md
index 9250c99b..9f76e45e 100644
--- a/docs/zh_cn/get_started/overview.md
+++ b/docs/zh_cn/get_started/overview.md
@@ -11,7 +11,7 @@ Lagent 是一个开源的 LLM 智能体框架，允许使用者快速将一个
 Lagent 包含三个主要模块：agents，llms 和 actions。
 
 - **agents** 实现了多种智能体，如 ReAct，AutoGPT。
-- **llms** 支持多种大语言模型，包括在 HuggingFace 上托管的开源模型（Llama-2, InterLM）及 GPT3.5/4 等闭源模型。
+- **llms** 支持多种大语言模型，包括在 HuggingFace 上托管的开源模型（Llama-2, InternLM）及 GPT3.5/4 等闭源模型。
 - **actions** 包含一系列工具，并提供工具执行器来统一管理。
 
 ## 如何使用
diff --git a/docs/zh_cn/get_started/quickstart.md b/docs/zh_cn/get_started/quickstart.md
index ae51f57b..e9a6f24b 100644
--- a/docs/zh_cn/get_started/quickstart.md
+++ b/docs/zh_cn/get_started/quickstart.md
@@ -15,7 +15,7 @@ from lagent.llms import GPTAPI
 # 初始化 LLM，你可能需要提供 API 密钥
 llm = GPTAPI(model_type='gpt-3.5-turbo', key=['Your OPENAI_API_KEY'])
 
-# 初始化 Goolge 搜索工具，你可能需要提供 API 密钥
+# 初始化 Google 搜索工具，你可能需要提供 API 密钥
 search_tool = GoogleSearch(api_key='Your SERPER_API_KEY')
 
 # 配置 ReWOO 智能体，创建聊天机器人

From 1a1f65bbc3c5a02861d2a4201fd44eec0cf40420 Mon Sep 17 00:00:00 2001
From: Jamie <jamiechoi1995@users.noreply.github.com>
Date: Thu, 17 Oct 2024 19:07:21 +0800
Subject: [PATCH 2/5] Update requirements (#245)

* Update optional.txt

* Update runtime.txt
---
 requirements/optional.txt | 2 +-
 requirements/runtime.txt  | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/requirements/optional.txt b/requirements/optional.txt
index 003acdb6..bcecba0a 100644
--- a/requirements/optional.txt
+++ b/requirements/optional.txt
@@ -4,5 +4,5 @@ pillow
 python-pptx
 timeout_decorator
 torch
-transformers>=4.34
+transformers>=4.34,<=4.40
 vllm>=0.3.3
diff --git a/requirements/runtime.txt b/requirements/runtime.txt
index 868aaa3e..fc30c3b1 100644
--- a/requirements/runtime.txt
+++ b/requirements/runtime.txt
@@ -2,7 +2,7 @@ arxiv
 distro
 filelock
 func_timeout
-griffe
+griffe<1.0
 json5
 jsonschema
 jupyter
@@ -13,4 +13,4 @@ streamlit
 tiktoken
 timeout-decorator
 typing-extensions
-duckduckgo_search==5.3.1b1
\ No newline at end of file
+duckduckgo_search==5.3.1b1

From 238aacf5e711f45cdbdfb26a6c2f4336fb64cd36 Mon Sep 17 00:00:00 2001
From: liukuikun <24622904+Harold-lkk@users.noreply.github.com>
Date: Thu, 17 Oct 2024 19:34:31 +0800
Subject: [PATCH 3/5] update requirement (#257)

---
 requirements/optional.txt | 2 +-
 requirements/runtime.txt  | 5 +++--
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/requirements/optional.txt b/requirements/optional.txt
index bcecba0a..0b095fa3 100644
--- a/requirements/optional.txt
+++ b/requirements/optional.txt
@@ -1,5 +1,5 @@
 google-search-results
-lmdeploy>=0.2.3
+lmdeploy<=0.5.3
 pillow
 python-pptx
 timeout_decorator
diff --git a/requirements/runtime.txt b/requirements/runtime.txt
index fc30c3b1..9fc7770a 100644
--- a/requirements/runtime.txt
+++ b/requirements/runtime.txt
@@ -1,5 +1,6 @@
 arxiv
 distro
+duckduckgo_search==5.3.1b1
 filelock
 func_timeout
 griffe<1.0
@@ -7,10 +8,10 @@ json5
 jsonschema
 jupyter
 jupyter_client
-phx-class-registry
+phx-class-registry==4.1.0
 requests
 streamlit
+termcolor
 tiktoken
 timeout-decorator
 typing-extensions
-duckduckgo_search==5.3.1b1

From f63e6ef93808db68856aba75b30bee81c7847dd4 Mon Sep 17 00:00:00 2001
From: Lyu Han <lvhan_028@163.com>
Date: Mon, 21 Oct 2024 13:16:24 +0800
Subject: [PATCH 4/5] Compatible with lmdeploy (#258)

* compatible with lmdeploy

* update

* update depends

* update

---------

Co-authored-by: liukuikun <24622904+Harold-lkk@users.noreply.github.com>
---
 lagent/llms/lmdeploy_wrapper.py | 51 +++++++++++++++++++++++++++++----
 requirements/optional.txt       |  2 +-
 2 files changed, 47 insertions(+), 6 deletions(-)

diff --git a/lagent/llms/lmdeploy_wrapper.py b/lagent/llms/lmdeploy_wrapper.py
index 4b11daff..0e9f2380 100644
--- a/lagent/llms/lmdeploy_wrapper.py
+++ b/lagent/llms/lmdeploy_wrapper.py
@@ -1,3 +1,5 @@
+import copy
+import logging
 from typing import List, Optional, Union
 
 from lagent.llms.base_llm import BaseModel
@@ -23,7 +25,12 @@ def __init__(self,
                  log_level: str = 'WARNING',
                  **kwargs):
         super().__init__(path=None, **kwargs)
-        from lmdeploy.serve.turbomind.chatbot import Chatbot, StatusCode
+        try:
+            from lmdeploy.serve.turbomind.chatbot import Chatbot, StatusCode
+        except Exception as e:
+            logging.error(f'{e}')
+            raise RuntimeError('DO NOT use turbomind.chatbot since it has '
+                               'been removed by lmdeploy since v0.5.2')
         self.state_map = {
             StatusCode.TRITON_STREAM_END: ModelStatusCode.END,
             StatusCode.TRITON_SERVER_ERR: ModelStatusCode.SERVER_ERR,
@@ -226,11 +233,32 @@ def __init__(self,
                  tp: int = 1,
                  pipeline_cfg=dict(),
                  **kwargs):
-
+        import lmdeploy
+        from lmdeploy import ChatTemplateConfig, TurbomindEngineConfig, pipeline, version_info
+
+        self.str_version = lmdeploy.__version__
+        self.version = version_info
+        self.do_sample = kwargs.pop('do_sample', None)
+        if self.do_sample is not None and self.version < (0, 6, 0):
+            raise RuntimeError(
+                '`do_sample` parameter is not supported by lmdeploy until '
+                f'v0.6.0, but currently using lmdeloy {self.str_version}')
         super().__init__(path=path, **kwargs)
-        from lmdeploy import pipeline
+        backend_config = copy.deepcopy(pipeline_cfg)
+        backend_config.update(tp=tp)
+        backend_config = {
+            k: v
+            for k, v in backend_config.items()
+            if hasattr(TurbomindEngineConfig, k)
+        }
+        backend_config = TurbomindEngineConfig(**backend_config)
+        chat_template_config = ChatTemplateConfig(
+            model_name=model_name) if model_name else None
         self.model = pipeline(
-            model_path=self.path, model_name=model_name, tp=tp, **pipeline_cfg)
+            model_path=self.path,
+            backend_config=backend_config,
+            chat_template_config=chat_template_config,
+            log_level='WARNING')
 
     def generate(self,
                  inputs: Union[str, List[str]],
@@ -249,13 +277,26 @@ def generate(self,
             (a list of/batched) text/chat completion
         """
         from lmdeploy.messages import GenerationConfig
-
         batched = True
         if isinstance(inputs, str):
             inputs = [inputs]
             batched = False
         prompt = inputs
+        do_sample = kwargs.pop('do_sample', None)
         gen_params = self.update_gen_params(**kwargs)
+
+        if do_sample is None:
+            do_sample = self.do_sample
+        if do_sample is not None and self.version < (0, 6, 0):
+            raise RuntimeError(
+                '`do_sample` parameter is not supported by lmdeploy until '
+                f'v0.6.0, but currently using lmdeloy {self.str_version}')
+        if self.version >= (0, 6, 0):
+            if do_sample is None:
+                do_sample = gen_params['top_k'] > 1 or gen_params[
+                    'temperature'] > 0
+            gen_params.update(do_sample=do_sample)
+
         gen_config = GenerationConfig(
             skip_special_tokens=skip_special_tokens, **gen_params)
         response = self.model.batch_infer(
diff --git a/requirements/optional.txt b/requirements/optional.txt
index 0b095fa3..75645dbe 100644
--- a/requirements/optional.txt
+++ b/requirements/optional.txt
@@ -1,5 +1,5 @@
 google-search-results
-lmdeploy<=0.5.3
+lmdeploy>=0.2.5
 pillow
 python-pptx
 timeout_decorator

From b41ade6a13f51769b4e6ce39142646aefcba0ca3 Mon Sep 17 00:00:00 2001
From: liukuikun <24622904+Harold-lkk@users.noreply.github.com>
Date: Mon, 21 Oct 2024 16:14:26 +0800
Subject: [PATCH 5/5] [Version] v0.2.4 (#261)

---
 lagent/version.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lagent/version.py b/lagent/version.py
index ff1c2005..92f5d587 100644
--- a/lagent/version.py
+++ b/lagent/version.py
@@ -1,5 +1,5 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-__version__ = '0.2.3'
+__version__ = '0.2.4'
 
 
 def parse_version_info(version_str: str, length: int = 4) -> tuple: