Skip to content

Commit 71d6d01

Browse files
mananroongtamananroongta
committed
rLLM and DeepSWE release
1 parent 6d45c5d commit 71d6d01

File tree

3 files changed

+47
-11
lines changed

3 files changed

+47
-11
lines changed

‎blog.html

Lines changed: 32 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -38,6 +38,38 @@ <h1>Archive</h1>
3838
<div class="year-label">2025
3939
<div class="posts-list">
4040

41+
<article class="blog-post">
42+
<div class="date">
43+
<h2>July<sup>1</sup></h2>
44+
</div>
45+
<div class="post-content">
46+
<h3>
47+
<a href="https://pretty-radio-b75.notion.site/rLLM-A-Framework-for-Post-Training-Language-Agents-21b81902c146819db63cd98a54ba5f31">
48+
rLLM: Reinforcement Learning for Language Agents
49+
</a>
50+
</h3>
51+
<p class="post-meta">
52+
Date: July 1, 2025 | Estimated Reading Time: 10 min | Author: Sijun Tan, Michael Luo, Colin Cai
53+
</p>
54+
</div>
55+
</article>
56+
57+
<article class="blog-post">
58+
<div class="date">
59+
<h2>July<sup>2</sup></h2>
60+
</div>
61+
<div class="post-content">
62+
<h3>
63+
<a href="https://pretty-radio-b75.notion.site/DeepSWE-Training-a-Fully-Open-sourced-State-of-the-Art-Coding-Agent-by-Scaling-RL-22281902c1468193aabbe9a8c59bbe33">
64+
DeepSWE: Training a Fully Open-sourced, State-of-the-Art Coding Agent by Scaling RL
65+
</a>
66+
</h3>
67+
<p class="post-meta">
68+
Date: July 1, 2025 | Estimated Reading Time: 20 min | Author: Agentica x Together AI
69+
</p>
70+
</div>
71+
</article>
72+
4173
<article class="blog-post">
4274
<div class="date">
4375
<h2>April<sup>1</sup></h2>
@@ -65,15 +97,6 @@ <h3><a href="https://pretty-radio-b75.notion.site/DeepScaleR-Surpassing-O1-Previ
6597
<p class="post-meta">Date: February 10, 2025 | Estimated Reading Time: 10 min | Author: Michael Luo, Sijun Tan</p>
6698
</div>
6799
</article>
68-
<!-- <article class="blog-post">
69-
<div class="date">
70-
<h2>July<sup>1</sup></h2>
71-
</div>
72-
<div class="post-content">
73-
<h3>I love Teedy, my favorite dog.</h3>
74-
<p class="post-meta">Date: January 1, 2025 | Estimated Reading Time: 1200 min | Author: Teedy</p>
75-
</div>
76-
</article> -->
77100
</div>
78101
</section>
79102

‎images/people/manan_roongta.jpg

-587 KB
Loading

‎index.html

Lines changed: 15 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -36,7 +36,7 @@ <h1 class="logo-text">Agentica</h1>
3636
<div class="hero-text">
3737
<h2>Welcome to the Agentica Project! 👋 </h2>
3838
<p class="main-paragraph">
39-
We are an open-source initiative to democratize reinforcement learning (RL) techniques and develop scalable systems for large language models (LLMs) and agents.
39+
We are a open-source initiative spawning from Berkeley Sky Computing Lab to democratize reinforcement learning (RL) techniques and develop scalable systems for large language models (LLMs) and agents.
4040
</p>
4141
<div class="social-icons">
4242
<a href="mailto:[email protected]">
@@ -57,12 +57,25 @@ <h2>Welcome to the Agentica Project! 👋 </h2>
5757

5858
<!-- ========== PROJECT SECTION ========== -->
5959
<section id="project">
60+
<a href="https://pretty-radio-b75.notion.site/rLLM-A-Framework-for-Post-Training-Language-Agents-21b81902c146819db63cd98a54ba5f31" rel="noopener noreferrer">
61+
<h3>rLLM: Reinforcement Learning for Language Agents</h3>
62+
<p class="main-paragraph">
63+
We release rLLM, an open-source framework for post-training language agents via reinforcement learning. With rLLM, you can easily build their custom agents and environments, train them with reinforcement learning, and deploy them for real-world workloads.
64+
</p>
65+
<div class="post-meta">Date: July 1, 2025 | Estimated Reading Time: 10 min | Author: Sijun Tan, Michael Luo, Colin Cai </div>
66+
</a>
67+
<a href="https://pretty-radio-b75.notion.site/DeepSWE-Training-a-Fully-Open-sourced-State-of-the-Art-Coding-Agent-by-Scaling-RL-22281902c1468193aabbe9a8c59bbe33" rel="noopener noreferrer">
68+
<h3>DeepSWE: Training a Fully Open-sourced, State-of-the-Art Coding Agent by Scaling RL</h3>
69+
<p class="main-paragraph">
70+
We release DeepSWE-Preview, a 32B software engineering agent (SWE) trained with purely RL that achieves 59% on SWEBench-Verified with test-time scaling,(42.2% Pass@1), topping the SWEBench leaderboard for open-weight models. </p>
71+
<div class="post-meta">Date: July 1, 2025 | Estimated Reading Time: 20 min | Author: Agentica x Together AI </div>
72+
</a>
6073
<a href="https://pretty-radio-b75.notion.site/DeepCoder-A-Fully-Open-Source-14B-Coder-at-O3-mini-Level-1cf81902c14680b3bee5eb349a512a51" rel="noopener noreferrer">
6174
<h3>DeepCoder: A Fully Open-Source 14B Coder at O3-mini Level</h3>
6275
<p class="main-paragraph">
6376
We release DeepCoder-14B-Preview, a code reasoning model finetuned from Deepseek-R1-Distilled-Qwen-14B via distributed RL. It achieves an impressive 60.6% Pass@1 accuracy on LiveCodeBench (+8% improvement), matching the performance of o3-mini-2025-01-031 (Low) and o1-2024-12-17 with just 14B parameters.
6477
</p>
65-
<div class="post-meta">Date: April 8, 2025 | Estimated Reading Time: 15 min | Author: Agentica x Together AI </div>
78+
<div class="post-meta">Date: July 1, 2025 | Estimated Reading Time: 15 min | Author: Agentica x Together AI </div>
6679
</a>
6780
<a href="https://pretty-radio-b75.notion.site/DeepScaleR-Surpassing-O1-Preview-with-a-1-5B-Model-by-Scaling-RL-19681902c1468005bed8ca303013a4e2" rel="noopener noreferrer">
6881
<h3>DeepScaleR: Surpassing O1-Preview with a 1.5B Model by Scaling RL</h3>

0 commit comments

Comments
 (0)