conflict

Author: abidlabs

_blog.yml

@@ -5921,6 +5921,17 @@
     - hub
     - inference
 
+- local: llama-guard-4
+  title: "Welcoming Llama Guard 4 on Hugging Face Hub"
+  author: merve
+  thumbnail: /blog/assets/llama-guard-4/thumbnail.png
+  date: April 29, 2025
+  tags:
+   - llama
+   - llm
+   - vision
+   - vlm
+
 - local: gradio-mcp 
   title: "How to Build an MCP Server with Gradio"
   author: abidlabs
@@ -5930,4 +5941,3 @@
     - gradio
     - tool
     - llm
-

lerobot-goes-to-driving-school.md

@@ -9,6 +9,16 @@ authors:
 
 ---
 
+# TL;DR
+
+A snapshot of [L2D](https://huggingface.co/datasets/yaak-ai/L2D), the world's largest self-driving dataset!
+- 90+ TeraBytes of multimodal data (5000+ hours of driving) from 30 cities in Germany.
+- 6 surrounding HD cameras + vehicle state (Speed/Heading/GPS/IMU)
+- Continuous (Gas/Brake/Steering) and discrete actions (Gear/Turn Signals)
+- OpenStreetMap [matched waypoints](#OpenStreetMap) from birds-eye-view.
+- Natural language instructions. F.ex ["When the light turns green, drive over the tram tracks and then through the roundabout"](https://huggingface.co/spaces/lerobot/visualize_dataset?dataset=yaak-ai%2FL2D&episode=82))
+- Expert (driving instructors) and student (learner drivers) policies
+
 # LeRobot goes to driving school
 
 State-of-the art [Vision Language Models](https://huggingface.co/blog/vlms) and Large Language Models are trained on open-source
@@ -249,8 +259,8 @@ information about the episodes. Each release **R1+** is a superset of the previo
 
 | HF | Nutron | Date | Episodes | Duration | Size | instructions  | task\_id | observation.state.route | suboptimal |
 | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
-| [R0](https://huggingface.co/datasets/yaak-ai/L2D) | [R0](https://nutron-sandbox.yaak.ai/collections/fcbb0dfd-40ae-4fd2-b023-7f300f35c5c7/300b7174-b6aa-4598-83e8-fc28cc5fcbe3/list) | March 2025 | 100 | 0.5+ hr | 9,5 GB | ☑️ |  |  |  |
-| R1 | R1 | April 2025 | 1K | 5+ hr | 95 GB | ☑️ |  |  |  |
+| [R0](https://huggingface.co/datasets/yaak-ai/L2D/tree/R0) | [R0](https://nutron-sandbox.yaak.ai/collections/fcbb0dfd-40ae-4fd2-b023-7f300f35c5c7/300b7174-b6aa-4598-83e8-fc28cc5fcbe3/search/list/session-logs?context=5s) | March 2025 | 100 | 0.5+ hr | 9,5 GB | ☑️ |  |  |  |
+| [R1](https://huggingface.co/datasets/yaak-ai/L2D) | [R1](https://nutron-sandbox.yaak.ai/collections/fcbb0dfd-40ae-4fd2-b023-7f300f35c5c7/1cb18573-f731-47b1-ae89-7ea2f026b8d0/search/list/session-logs?context=5s) | April 2025 | 1K | 5+ hr | 95 GB | ☑️ |  |  |  |
 | R2 | R2 | May 2025 | 10K | 50+ hr | 1 TB | ☑️ | ☑️ | ☑️ | ☑️ |
 | R3 | R3 | June 2025 | 100K | 500+ hr | 10 TB | ☑️ | ☑️ | ☑️ | ☑️ |
 | R4 | R4 | July 2025 | 1M | 5000+ hr | 90 TB | ☑️ | ☑️ | ☑️ | ☑️ |

open-deep-research.md

@@ -28,14 +28,17 @@ The clock is ticking, let’s go! ⏱️
 
 ## Table of Contents
 
-- [What are Agent frameworks and why they matter?](#what-are-agent-frameworks-and-why-they-matter)
-- [The GAIA benchmark](#the-gaia-benchmark)
-- [Building an open Deep Research](#building-an-open-deep-research)
-  - [Using a CodeAgent](#using-a-codeagent)
-  - [Making the right tools 🛠️](#making-the-right-tools-🛠️)
-- [Results 🏅](#results-🏅)
-- [Community reproductions](#community-reproductions)
-- [Most important next steps](#most-important-next-steps)
+- [Open-source DeepResearch – Freeing our search agents](#open-source-deepresearch--freeing-our-search-agents)
+  - [TLDR](#tldr)
+  - [Table of Contents](#table-of-contents)
+  - [What are Agent frameworks and why they matter?](#what-are-agent-frameworks-and-why-they-matter)
+  - [The GAIA benchmark](#the-gaia-benchmark)
+  - [Building an open Deep Research](#building-an-open-deep-research)
+    - [Using a CodeAgent](#using-a-codeagent)
+    - [Making the right tools 🛠️](#making-the-right-tools-️)
+  - [Results 🏅](#results-)
+  - [Community Reproductions](#community-reproductions)
+  - [Most important next steps](#most-important-next-steps)
 
 
 ## What are Agent frameworks and why they matter?
@@ -112,17 +115,18 @@ From building `smolagents` we can also cite a notable additional advantage, whic
 
 Now we need to provide the agent with the right set of tools. 
 
-**1.** A web browser. While a fully fledged web browser interaction like [Operator](https://openai.com/index/introducing-operator/) will be needed to reach full performance, we started with an extremely simple text-based web browser for now for our first proof-of-concept. You can find the code [here](https://github.com/huggingface/smolagents/blob/gaia-submission-r1/examples/open_deep_research/scripts/text_web_browser.py)
+**1.** A web browser. While a fully fledged web browser interaction like [Operator](https://openai.com/index/introducing-operator/) will be needed to reach full performance, we started with an extremely simple text-based web browser for now for our first proof-of-concept. You can find the code [here](https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research/scripts/text_web_browser.py)
 
-**2.** A simple text inspector, to be able to **read a bunch of text file format**, find it [here](https://github.com/huggingface/smolagents/blob/gaia-submission-r1/examples/open_deep_research/scripts/text_inspector_tool.py).
+
+**2.** A simple text inspector, to be able to **read a bunch of text file format**, find it [here](https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research/scripts/text_inspector_tool.py).
 
 These tools were taken from the excellent [Magentic-One](https://www.microsoft.com/en-us/research/articles/magentic-one-a-generalist-multi-agent-system-for-solving-complex-tasks/) agent by Microsoft Research, kudos to them! We didn’t change them much, as our goal was to get as high a performance as we can with the lowest complexity possible.
 
 Here is a short roadmap of improvements which we feel would really improve these tools’ performance (feel free to open a PR and contribute!):
 
 - extending the number of file formats which can be read.
 - proposing a more fine-grained handling of files.
-- replacing the web browser with a vision-based one, which we’ve started doing [here](https://github.com/huggingface/smolagents/blob/gaia-submission-r1/src/smolagents/vision_web_browser.py).
+- replacing the web browser with a vision-based one, which we’ve started doing [here](https://github.com/huggingface/smolagents/tree/main/src/smolagents/vision_web_browser.py).
 
 ## Results 🏅
 
@@ -169,6 +173,6 @@ So we’re tackling that next! In a more general problem: we’re going to build
 
 We’re also [hiring a full time engineer](https://apply.workable.com/huggingface/j/AF1D4E3FEB/) to help us work on this and more, apply if you’re interested 🙂
 
-- To get started with Open Deep Research, try the examples [here](https://github.com/huggingface/smolagents/tree/gaia-submission-r1/examples/open_deep_research).
+- To get started with Open Deep Research, try the examples [here](https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research).
 - Check the [smolagents](https://github.com/huggingface/smolagents) repo.
 - Read more about smolagents [docs](https://huggingface.co/docs/smolagents/index), [introduction blog post](https://huggingface.co/blog/smolagents).