Merge pull request #5 from huggingface/main

Update from HF.

Author: joaopauloschuler

.github/workflows/build_documentation.yml

@@ -11,6 +11,7 @@ on:
       - 'docs/source/**'
       - 'assets/**'
       - '.github/workflows/doc-build.yml'
+      - 'pyproject.toml'
 
 jobs:
    build:

docs/source/en/examples/multiagents.md

@@ -169,7 +169,7 @@ manager_agent = CodeAgent(
 That's all! Now let's run our system! We select a question that requires both some calculation and research:
 
 ```py
-answer = manager_agent.run("If LLM trainings continue to scale up at the current rhythm until 2030, what would be the electric power in GW required to power the biggest training runs by 2030? What does that correspond to, compared to some contries? Please provide a source for any number used.")
+answer = manager_agent.run("If LLM training continues to scale up at the current rhythm until 2030, what would be the electric power in GW required to power the biggest training runs by 2030? What would that correspond to, compared to some countries? Please provide a source for any numbers used.")
 ```
 
 We get this report as the answer:

docs/source/en/examples/rag.md

@@ -29,7 +29,7 @@ This agent will: ✅ Formulate the query itself and ✅ Critique to re-retrieve
 
 So it should naively recover some advanced RAG techniques!
 - Instead of directly using the user query as the reference in semantic search, the agent formulates itself a reference sentence that can be closer to the targeted documents, as in [HyDE](https://huggingface.co/papers/2212.10496).
-The agent can the generated snippets and re-retrieve if needed, as in [Self-Query](https://docs.llamaindex.ai/en/stable/examples/evaluation/RetryQuery/).
+The agent can use the generated snippets and re-retrieve if needed, as in [Self-Query](https://docs.llamaindex.ai/en/stable/examples/evaluation/RetryQuery/).
 
 Let's build this system. 🛠️
 

docs/source/en/tutorials/building_good_agents.md

@@ -233,9 +233,9 @@ Here are the rules you should always follow to solve your task:
 Now Begin! If you solve the task correctly, you will receive a reward of $1,000,000.
 ```
 
-As yo can see, there are placeholders like `"{{tool_descriptions}}"`: these will be used upon agent initialization to insert certain automatically generated descriptions of tools or managed agents.
+As you can see, there are placeholders like `"{{tool_descriptions}}"`: these will be used upon agent initialization to insert certain automatically generated descriptions of tools or managed agents.
 
-So while you can overwrite this system prompt template by passing your custom prompt as an argument to the `system_prompt` parameter, your new system promptmust contain the following placeholders:
+So while you can overwrite this system prompt template by passing your custom prompt as an argument to the `system_prompt` parameter, your new system prompt must contain the following placeholders:
 - `"{{tool_descriptions}}"` to insert tool descriptions.
 - `"{{managed_agents_description}}"` to insert the description for managed agents if there are any.
 - For `CodeAgent` only: `"{{authorized_imports}}"` to insert the list of authorized imports.

docs/source/en/tutorials/secure_code_execution.md

@@ -47,7 +47,7 @@ This interpreter is designed for security by:
  - Capping the number of operations to prevent infinite loops and resource bloating.
  - Will not perform any operation that's not pre-defined.
 
-Wev'e used this on many use cases, without ever observing any damage to the environment. 
+We've used this on many use cases, without ever observing any damage to the environment. 
 
 However this solution is not watertight: one could imagine occasions where LLMs fine-tuned for malignant actions could still hurt your environment. For instance if you've allowed an innocuous package like `Pillow` to process images, the LLM could generate thousands of saves of images to bloat your hard drive.
 It's certainly not likely if you've chosen the LLM engine yourself, but it could happen.
@@ -79,4 +79,4 @@ agent = CodeAgent(
 agent.run("What was Abraham Lincoln's preferred pet?")
 ```
 
-E2B code execution is not compatible with multi-agents at the moment - because having an agent call in a code blob that should be executed remotely is a mess. But we're working on adding it!
+E2B code execution is not compatible with multi-agents at the moment - because having an agent call in a code blob that should be executed remotely is a mess. But we're working on adding it!

docs/source/en/tutorials/tools.md

@@ -89,8 +89,8 @@ model_downloads_tool.push_to_hub("{your_username}/hf-model-downloads", token="<Y
 ```
 
 For the push to Hub to work, your tool will need to respect some rules:
-- All method are self-contained, e.g. use variables that come either from their args.
-- As per the above point, **all imports should be defined directky within the tool's functions**, else you will get an error when trying to call [`~Tool.save`] or [`~Tool.push_to_hub`] with your custom tool.
+- All methods are self-contained, e.g. use variables that come either from their args.
+- As per the above point, **all imports should be defined directly within the tool's functions**, else you will get an error when trying to call [`~Tool.save`] or [`~Tool.push_to_hub`] with your custom tool.
 - If you subclass the `__init__` method, you can give it no other argument than `self`. This is because arguments set during a specific tool instance's initialization are hard to track, which prevents from sharing them properly to the hub. And anyway, the idea of making a specific class is that you can already set class attributes for anything you need to hard-code (just set `your_variable=(...)` directly under the `class YourTool(Tool):` line). And of course you can still create a class attribute anywhere in your code by assigning stuff to `self.your_variable`.
 
 
@@ -210,7 +210,7 @@ Just make sure the new tool follows the same API as the replaced tool or adapt t
 ### Use a collection of tools
 
 You can leverage tool collections by using the ToolCollection object, with the slug of the collection you want to use.
-Then pass them as a list to initialize you agent, and start using them!
+Then pass them as a list to initialize your agent, and start using them!
 
 ```py
 from smolagents import ToolCollection, CodeAgent
@@ -224,4 +224,4 @@ agent = CodeAgent(tools=[*image_tool_collection.tools], model=model, add_base_to
 agent.run("Please draw me a picture of rivers and lakes.")
 ```
 
-To speed up the start, tools are loaded only if called by the agent.
+To speed up the start, tools are loaded only if called by the agent.

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "smolagents"
-version = "1.1.0.dev0"
+version = "1.2.0.dev0"
 description = "🤗 smolagents: a barebones library for agents. Agents write python code to call tools or orchestrate other agents."
 authors = [
   { name="Aymeric Roucher", email="[email protected]" }, { name="Thomas Wolf"},

src/smolagents/__init__.py

@@ -14,7 +14,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-__version__ = "1.1.0.dev0"
+__version__ = "1.2.0.dev0"
 
 from typing import TYPE_CHECKING
 

src/smolagents/default_tools.py

@@ -146,12 +146,11 @@ def forward(self, question):
 
 class DuckDuckGoSearchTool(Tool):
     name = "web_search"
-    description = """Performs a duckduckgo web search based on your query (think a Google search) then returns the top search results as a list of dict elements.
-    Each result has keys 'title', 'href' and 'body'."""
+    description = """Performs a duckduckgo web search based on your query (think a Google search) then returns the top search results."""
     inputs = {
         "query": {"type": "string", "description": "The search query to perform."}
     }
-    output_type = "any"
+    output_type = "string"
 
     def __init__(self, *args, max_results=10, **kwargs):
         super().__init__(*args, **kwargs)

src/smolagents/gradio_ui.py

@@ -23,7 +23,7 @@
 def pull_messages_from_step(step_log: AgentStep, test_mode: bool = True):
     """Extract ChatMessage objects from agent steps"""
     if isinstance(step_log, ActionStep):
-        yield gr.ChatMessage(role="assistant", content=step_log.llm_output)
+        yield gr.ChatMessage(role="assistant", content=step_log.llm_output or "")
         if step_log.tool_call is not None:
             used_code = step_log.tool_call.name == "code interpreter"
             content = step_log.tool_call.arguments
@@ -35,9 +35,7 @@ def pull_messages_from_step(step_log: AgentStep, test_mode: bool = True):
                 content=str(content),
             )
         if step_log.observations is not None:
-            yield gr.ChatMessage(
-                role="assistant", content=f"```\n{step_log.observations}\n```"
-            )
+            yield gr.ChatMessage(role="assistant", content=step_log.observations)
         if step_log.error is not None:
             yield gr.ChatMessage(
                 role="assistant",
@@ -65,7 +63,7 @@ def stream_to_gradio(
     if isinstance(final_answer, AgentText):
         yield gr.ChatMessage(
             role="assistant",
-            content=f"**Final answer:**\n```\n{final_answer.to_string()}\n```",
+            content=f"**Final answer:**\n{final_answer.to_string()}\n",
         )
     elif isinstance(final_answer, AgentImage):
         yield gr.ChatMessage(