Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
| 1 |
You can download the model then run the inference scipts in https://github.com/Alibaba-NLP/WebAgent.
|
| 2 |
|
| 3 |
- **WebSailor** is a complete post-training methodology designed to teach LLM agents sophisticated reasoning for complex web navigation and information-seeking tasks. It addresses the challenge of extreme uncertainty in vast information landscapes, a capability where previous open-source models lagged behind proprietary systems.
|
|
@@ -6,4 +9,4 @@ You can download the model then run the inference scipts in https://github.com/A
|
|
| 6 |
|
| 7 |
- Our training process begins by generating expert trajectories and then reconstructing the reasoning to create concise, action-oriented supervision signals, avoiding the stylistic and verbosity issues of teacher models. The agent is first given a "cold start" using rejection sampling fine-tuning (RFT) on a small set of high-quality examples to establish a baseline capability. This is followed by an efficient agentic reinforcement learning stage using our **Duplicating Sampling Policy Optimization (DUPO)** algorithm, which refines the agent's exploratory strategies.
|
| 8 |
|
| 9 |
-
- WebSailor establishes a **new state-of-the-art for open-source agents**, achieving outstanding results on difficult benchmarks like BrowseComp-en and BrowseComp-zh. Notably, our smaller models like WebSailor-7B outperform agents built on much larger backbones, highlighting the efficacy of our training paradigm. Ultimately, WebSailor closes the performance gap to proprietary systems, achieving results on par with agents like Doubao-Search.
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
---
|
| 4 |
You can download the model then run the inference scipts in https://github.com/Alibaba-NLP/WebAgent.
|
| 5 |
|
| 6 |
- **WebSailor** is a complete post-training methodology designed to teach LLM agents sophisticated reasoning for complex web navigation and information-seeking tasks. It addresses the challenge of extreme uncertainty in vast information landscapes, a capability where previous open-source models lagged behind proprietary systems.
|
|
|
|
| 9 |
|
| 10 |
- Our training process begins by generating expert trajectories and then reconstructing the reasoning to create concise, action-oriented supervision signals, avoiding the stylistic and verbosity issues of teacher models. The agent is first given a "cold start" using rejection sampling fine-tuning (RFT) on a small set of high-quality examples to establish a baseline capability. This is followed by an efficient agentic reinforcement learning stage using our **Duplicating Sampling Policy Optimization (DUPO)** algorithm, which refines the agent's exploratory strategies.
|
| 11 |
|
| 12 |
+
- WebSailor establishes a **new state-of-the-art for open-source agents**, achieving outstanding results on difficult benchmarks like BrowseComp-en and BrowseComp-zh. Notably, our smaller models like WebSailor-7B outperform agents built on much larger backbones, highlighting the efficacy of our training paradigm. Ultimately, WebSailor closes the performance gap to proprietary systems, achieving results on par with agents like Doubao-Search.
|