混元大模型算法研究员-模型评测
Tencent
混元大模型算法研究员-模型评测
Job Description
混元大模型算法研究员-模型评测
📋 Job Overview
The role of Algorithm Researcher for Model Evaluation in Tencent's Hunyuan Large Model focuses on advancing the evaluation paradigms for large language models. Responsibilities include designing evaluation systems, building data pipelines, developing analysis platforms, and establishing diagnostic systems for pre-trained models. This position aims to scientifically explore and define the cognitive boundaries of AI models in complex scenarios.
📍 Location: Beijing, China
🏢 Business Unit: TEG
📄 Full Description
1.定义下一代评测范式：设计并建立一套能系统性地衡量大模型在开放、复杂场景下的真实能力评测体系（包括不局限于Agent、长文等），以科学地探索并定义模型的认知边界；
2.构建科学严谨的评测数据体系：能将对模型能力的理解转化为科学、严谨的数据设计与标注规范，构建自动化评测数据合成管线，为整个评测体系提供高质量的数据基石；
3.研发评估分析平台：研发专家级“Judge Model”与深度诊断框架的智能化评估分析平台，以实现高效、精准的自动化评估与归因；
4.建立预训练基座模型的深度诊断系统：分析预训练模型在知识内化、推理能力等层面的核心特性与能力边界，形成对模型的科学认知，为模型的预训练数据配比、算法优化等关键决策提供指导。
2.构建科学严谨的评测数据体系：能将对模型能力的理解转化为科学、严谨的数据设计与标注规范，构建自动化评测数据合成管线，为整个评测体系提供高质量的数据基石；
3.研发评估分析平台：研发专家级“Judge Model”与深度诊断框架的智能化评估分析平台，以实现高效、精准的自动化评估与归因；
4.建立预训练基座模型的深度诊断系统：分析预训练模型在知识内化、推理能力等层面的核心特性与能力边界，形成对模型的科学认知，为模型的预训练数据配比、算法优化等关键决策提供指导。
🎯 Key Responsibilities
- Define next-generation evaluation paradigms: Design and establish a systematic evaluation system to measure the true capabilities of large models in open, complex scenarios (including but not limited to Agent, long text, etc.), scientifically explore and define the cognitive boundaries of models.
- Build a scientific and rigorous evaluation data system: Translate understanding of model capabilities into scientific, rigorous data design and annotation standards, construct automated evaluation data synthesis pipelines, providing a high-quality data foundation for the entire evaluation system.
- Develop evaluation analysis platform: Research and develop an intelligent evaluation analysis platform with expert-level 'Judge Model' and deep diagnostic frameworks to achieve efficient, precise automated evaluation and attribution.
- Establish deep diagnostic system for pre-trained base models: Analyze core characteristics and capability boundaries of pre-trained models in aspects such as knowledge internalization and reasoning abilities, form scientific understanding of models, and provide guidance for key decisions like pre-training data allocation and algorithm optimization.
Locations
- Beijing, China
Salary
Estimated Salary Rangemedium confidence
300,000 - 600,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Responsibilities
- Define next-generation evaluation paradigms: Design and establish a systematic evaluation system to measure the true capabilities of large models in open, complex scenarios (including but not limited to Agent, long text, etc.), scientifically explore and define the cognitive boundaries of models.
- Build a scientific and rigorous evaluation data system: Translate understanding of model capabilities into scientific, rigorous data design and annotation standards, construct automated evaluation data synthesis pipelines, providing a high-quality data foundation for the entire evaluation system.
- Develop evaluation analysis platform: Research and develop an intelligent evaluation analysis platform with expert-level 'Judge Model' and deep diagnostic frameworks to achieve efficient, precise automated evaluation and attribution.
- Establish deep diagnostic system for pre-trained base models: Analyze core characteristics and capability boundaries of pre-trained models in aspects such as knowledge internalization and reasoning abilities, form scientific understanding of models, and provide guidance for key decisions like pre-training data allocation and algorithm optimization.
Tags & Categories
Answer 10 quick questions to check your fit for 混元大模型算法研究员-模型评测 @ Tencent.
10 Questions
~2 Minutes
Instant Score
Related Books and Jobs
混元大模型算法研究员-模型评测
Tencent
混元大模型算法研究员-模型评测
Job Description
混元大模型算法研究员-模型评测
📋 Job Overview
The role of Algorithm Researcher for Model Evaluation in Tencent's Hunyuan Large Model focuses on advancing the evaluation paradigms for large language models. Responsibilities include designing evaluation systems, building data pipelines, developing analysis platforms, and establishing diagnostic systems for pre-trained models. This position aims to scientifically explore and define the cognitive boundaries of AI models in complex scenarios.
📍 Location: Beijing, China
🏢 Business Unit: TEG
📄 Full Description
1.定义下一代评测范式：设计并建立一套能系统性地衡量大模型在开放、复杂场景下的真实能力评测体系（包括不局限于Agent、长文等），以科学地探索并定义模型的认知边界；
2.构建科学严谨的评测数据体系：能将对模型能力的理解转化为科学、严谨的数据设计与标注规范，构建自动化评测数据合成管线，为整个评测体系提供高质量的数据基石；
3.研发评估分析平台：研发专家级“Judge Model”与深度诊断框架的智能化评估分析平台，以实现高效、精准的自动化评估与归因；
4.建立预训练基座模型的深度诊断系统：分析预训练模型在知识内化、推理能力等层面的核心特性与能力边界，形成对模型的科学认知，为模型的预训练数据配比、算法优化等关键决策提供指导。
2.构建科学严谨的评测数据体系：能将对模型能力的理解转化为科学、严谨的数据设计与标注规范，构建自动化评测数据合成管线，为整个评测体系提供高质量的数据基石；
3.研发评估分析平台：研发专家级“Judge Model”与深度诊断框架的智能化评估分析平台，以实现高效、精准的自动化评估与归因；
4.建立预训练基座模型的深度诊断系统：分析预训练模型在知识内化、推理能力等层面的核心特性与能力边界，形成对模型的科学认知，为模型的预训练数据配比、算法优化等关键决策提供指导。
🎯 Key Responsibilities
- Define next-generation evaluation paradigms: Design and establish a systematic evaluation system to measure the true capabilities of large models in open, complex scenarios (including but not limited to Agent, long text, etc.), scientifically explore and define the cognitive boundaries of models.
- Build a scientific and rigorous evaluation data system: Translate understanding of model capabilities into scientific, rigorous data design and annotation standards, construct automated evaluation data synthesis pipelines, providing a high-quality data foundation for the entire evaluation system.
- Develop evaluation analysis platform: Research and develop an intelligent evaluation analysis platform with expert-level 'Judge Model' and deep diagnostic frameworks to achieve efficient, precise automated evaluation and attribution.
- Establish deep diagnostic system for pre-trained base models: Analyze core characteristics and capability boundaries of pre-trained models in aspects such as knowledge internalization and reasoning abilities, form scientific understanding of models, and provide guidance for key decisions like pre-training data allocation and algorithm optimization.
Locations
- Beijing, China
Salary
Estimated Salary Rangemedium confidence
300,000 - 600,000 CNY / yearly
Source: ai estimated
* This is an estimated range based on market data and may vary based on experience and qualifications.
Responsibilities
- Define next-generation evaluation paradigms: Design and establish a systematic evaluation system to measure the true capabilities of large models in open, complex scenarios (including but not limited to Agent, long text, etc.), scientifically explore and define the cognitive boundaries of models.
- Build a scientific and rigorous evaluation data system: Translate understanding of model capabilities into scientific, rigorous data design and annotation standards, construct automated evaluation data synthesis pipelines, providing a high-quality data foundation for the entire evaluation system.
- Develop evaluation analysis platform: Research and develop an intelligent evaluation analysis platform with expert-level 'Judge Model' and deep diagnostic frameworks to achieve efficient, precise automated evaluation and attribution.
- Establish deep diagnostic system for pre-trained base models: Analyze core characteristics and capability boundaries of pre-trained models in aspects such as knowledge internalization and reasoning abilities, form scientific understanding of models, and provide guidance for key decisions like pre-training data allocation and algorithm optimization.
Tags & Categories
Answer 10 quick questions to check your fit for 混元大模型算法研究员-模型评测 @ Tencent.
10 Questions
~2 Minutes
Instant Score
Related Books and Jobs