Deep Principle有发布氢-1.0，是化学，材料科学和生命科学的开放式科学模型。 它由训练后Z.ai的GLM-5.3在域数据，实验室记录和agent发现轨迹上建造。 对于开发者来说，它提供报告一个Frontier Science-Research胜过GPT-6-Astra的开放权重。 麻省理工学院许可下的重量船
TL；数据
- 大小：753B参数(MoE, ~40B活动于GLM-5.3基数),1M-token上下文，128K最大输出。
- 运行时间：8GPU带SGLang(测试);NVFP4为NVIDIA B200/B300,MXFP4为AMD MI350X/MI355X.
- 性能：比GPT-6-Astra关于前沿科学-研究；大致比对GPQA-Diamond上的Kimi-K3.
- 最佳：前沿科学研究38.3%,GPT-6-Astra(自报)33.3%。
- 最糟糕的是：在NMR结构说明上，拖车GPT-6-Astra(最大思考)得3.33分。
什么是氢-10.
Hydrogen-1.0是来自深原原则的科研助理模型，是一家以杭州为基地的AI-换材料公司。 模型卡将其描述为用于科学解决问题的GLM-5.3衍生物。 它的目标是问题框架、试验规划、结果分析和无限制的发现。 Deep Principle团队表示，专有模型产生了最近的AI科学成果，但其发现能力并不开放。 氢-1.0是它尝试一种可复制的，可检验的替代品。
氢-1.0是如何训练的？
管道结合了受监督的微调(SFT)与强化学习(RL). 数据来自4个来源：
- 领域知识：化学、材料科学和生命科学数据。
- 研究设想：深原则科学家针对具体任务的反馈。
- 实验室记录：通过米拉获取的全周期电子实验室笔记本记录。
- 发现轨迹：选择和反馈的顺序，包括失败的尝试。
基础设施以威震天和SGLang为基础，跨越NVIDIA CUDA和AMD ROCm. 团队保持动态稀疏的注意力因果跨上下文平行的硬。 Muon优化器状态在CPU,GPU和磁盘每个后端之间移动。 RL使用可回收的推出，每个令牌都保留其原有的行为日志概率以进行重要性校正。 一个布局适配器将训练检查站转换为BF16和128×128区块式FP8推断权重。 完整的数据集和配方不是发布。
基准测试上的氢-1.0表现如何？
关于公开的基准测试，深原则报告Hydrogen-1.0的成绩超过了GPT-6-Astra关于前沿科学-研究。 在GPQA-Diamond上，这在GLM-5.3上有所改进
内部研究任务 evals 将其与 GPT-6-Astra 进行最大思考：
- 科学计算：氢-1.0领先5.13个百分点。
- 前进反应预测：氢-1.0领先4.76个百分点。
- NMR结构说明：GPT-6-Astra领先3.33分。
在过渡金属复合设计中，模型在5轮内追逐更大的HOMO-LUMO缺口。 到第5轮时，氢-1.0达到了平均最远差4.318eV. GPT-6-Astra达到了3.689eV,Kimi-K3达到了3.353eV. 在117个实验规划问题中，超过Kimi-K3. GPT-6-Astra和Opus-5.5的氢1.0得分在拒绝超过50%的请求后被排除在外。
在真正的实验室工作流程中能做什么？
技术帖显示4演示。 氢-1.0写作Python来构建聚乙烯熔融，并运行GROMACS. 它检测到一个地形错误，修复它，并分析修正的40ns轨迹。 它通过30种可信的反应，形成一个由27个结构组成的爱尔兰-克莱森反应网络。 它从1H NMR的痕量中将2-溴甲苯的7种候选单成型产物排在首位。 也为12年的铃木耦合研究撰写了JSON协议。 虚拟的实验室编译步骤在处决前捕捉到没有新鲜提示的欲望。
如何部署氢-1.0?
重量在Hugging Face和ModelScope上。 Deep Principal team在8 GPU上使用8. 的拉伸平行法测试了与SGLang一起服务的BF16. NVFP4针对NVIDIA B200和B300. MXFP4针对AMD Instinct MI350X和MI355X. 服务器将OpenAI兼容的API与推理 effort设置曝光。
氢-1.0 vs GLM-5.3, Kimi K3和GPT-6-Astra
关键外卖
- Hydrogen-1.0是麻省理工学院颁发的许可证，753B GLM-5.3型科学后列车。
- 它报告38.3%的前沿科学研究，高于GPT-6-Astra的33.3%。
- 训练新增实验室记录和发现轨迹，而不仅仅是论文。
- 它仍然跟踪GPT-6-Astra 在NMR结构澄清。
- NVFP4和MXFP4搭建目标布莱克威尔和MI350级GPU.
请检查access-date=中的日期值 (帮助) Technical blog, Model grates on Hugging Face and Usage 指出。 并加入我们的通讯。 等等你在电报里吗现在你也可以加入我们的电报。
需要和我们合作推广您的 GitHub Repo OR Hugging Face Page OR Product Release OR Webinar 等吗？联系我们
Asif Razzaq是Marktechpost AI Media Inc.的首席执行官。 作为一个有远见的企业家和工程师，阿西夫致力于利用人工智能的潜力促进社会福祉。 他最近的努力是人工智能媒体平台的发布，该平台以深入报道机器学习和深层学习新闻而著称，这些新闻在技术上既可靠又为广大观众所容易理解。 该平台每月有超过200万次观看，表明其受观众欢迎。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Deep Principle has released Hydrogen-1.0, an open-weight scientific model for chemistry, materials science and life sciences. It is built by post-training Z.ai’s GLM-5.3 on domain data, laboratory records and agent discovery trajectories. For developers, it offers open weights that report a FrontierScience-Research win over GPT-6-Astra. The weights ship under the MIT License.
TL;DR
- Size: 753B parameters (MoE, ~40B active per the GLM-5.3 base), 1M-token context, 128K max output.
- Runs on: 8 GPUs with SGLang (tested); NVFP4 for NVIDIA B200/B300, MXFP4 for AMD MI350X/MI355X.
- Performance: Beats GPT-6-Astra on FrontierScience-Research; roughly matches Kimi-K3 on GPQA-Diamond.
- Best: 38.3% on FrontierScience-Research vs 33.3% for GPT-6-Astra (self-reported).
- Worst: Trails GPT-6-Astra (max thinking) by 3.33 points on NMR structure elucidation.
What is Hydrogen-1.0?
Hydrogen-1.0 is a scientific research assistant model from Deep Principle, a Hangzhou-based AI-for-materials company. The model card describes it as a GLM-5.3 derivative for scientific problem solving. It targets question framing, experiment planning, results analysis and open-ended discovery. Deep Principle team states that proprietary models produced recent AI science results, but their discovery capabilities are not open. Hydrogen-1.0 is its attempt at a reproducible, inspectable alternative.
How was Hydrogen-1.0 trained?
The pipeline combines supervised fine-tuning (SFT) with reinforcement learning (RL). Data comes from 4 sources:
- Domain knowledge: chemistry, materials science and life sciences data.
- Research scenarios: task-specific feedback from Deep Principle scientists.
- Laboratory records: full-cycle electronic lab notebook records captured through Mira.
- Discovery trajectories: sequences of choices and feedback, including failed attempts.
Infrastructure builds on Megatron and SGLang across NVIDIA CUDA and AMD ROCm. The team keeps dynamic sparse attention causal across context-parallel shards. Muon optimizer state is moved between CPU, GPU and disk per backend. RL uses resumable rollouts, and each token keeps its original behavior log-probability for importance correction. A layout adapter converts training checkpoints into BF16 and 128×128 blockwise FP8 inference weights. Full datasets and recipes are not released.
How does Hydrogen-1.0 perform on benchmarks?
On public benchmarks, Deep Principle reports Hydrogen-1.0 outperforms GPT-6-Astra on FrontierScience-Research. The tweet thread from one of the founders’ report 93.0% on GPQA-Diamond. That improves on GLM-5.3 and roughly matches Kimi-K3, per the blog.
Internal research-task evals compare it with GPT-6-Astra at maximum thinking:
- Scientific computing: Hydrogen-1.0 leads by 5.13 percentage points.
- Forward reaction prediction: Hydrogen-1.0 leads by 4.76 percentage points.
- NMR structure elucidation: GPT-6-Astra leads by 3.33 points.
In transition metal complex design, models chased a larger HOMO-LUMO gap over 5 rounds. By round 5, Hydrogen-1.0 reached a mean best-so-far gap of 4.318 eV. GPT-6-Astra reached 3.689 eV and Kimi-K3 reached 3.353 eV. On 117 experiment-planning questions, Hydrogen-1.0 scored above Kimi-K3. GPT-6-Astra and Opus-5.5 were excluded after refusing more than 50% of requests.
What can it do in a real lab workflow?
The technical post shows 4 demos. Hydrogen-1.0 writes Python to build a polyethylene melt and runs GROMACS. It detects a topology error, repairs it, and analyzes the corrected 40 ns trajectory. It grows an Ireland-Claisen reaction network of 27 structures through 30 plausible reactions. It ranks 7 candidate monoformylation products of 2-bromonaphthalene from a 1H NMR trace. It also writes a JSON protocol for a 12-vial Suzuki coupling study. A virtual lab compilation step catches an aspiration without a fresh tip before execution.
How do you deploy Hydrogen-1.0?
Weights are on Hugging Face and ModelScope. Deep Principle team tested BF16 serving with SGLang on 8 GPUs using tensor parallelism of 8. NVFP4 targets NVIDIA B200 and B300. MXFP4 targets AMD Instinct MI350X and MI355X. The server exposes an OpenAI-compatible API with a reasoning_effort setting.
Hydrogen-1.0 vs GLM-5.3, Kimi K3 and GPT-6-Astra
Key Takeaways
- Hydrogen-1.0 is an MIT-licensed, 753B GLM-5.3 post-train for science.
- It reports 38.3% on FrontierScience-Research, above GPT-6-Astra’s 33.3%.
- Training adds lab records and discovery trajectories, not just papers.
- It still trails GPT-6-Astra on NMR structure elucidation.
- NVFP4 and MXFP4 builds target Blackwell and MI350-class GPUs.
Check out the Technical blog, Model weights on Hugging Face and Usage notes. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
Asif Razzaq is the CEO of Marktechpost AI Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.