JobbSafariLediga jobbMaster's Thesis: Attention-Based Hallucination Probability Estimation in Large Language Models
Master's Thesis: Attention-Based Hallucination Probability Estimation in Large Language Models
Ericsson ABNy
Sammanfattning
This thesis opportunity focuses on evaluating and developing methods to predict hallucinations in large language models (LLMs), particularly in telecommunications applications. Conducted in Stockholm, Sweden, under the supervision of Konrad Nowosadko, the project aims to create a risk-estimation framework for hallucination detection. The role is suitable for final-year Master's students in relevant fields, starting in January 2027.Det här erbjuder vi
Opportunity to work on innovative research in a cutting-edge field.Collaborate with a diverse team of innovators.Engagement in a supportive and inclusive work environment.
Ansök senast: Öppet tillsvidare
Publicerad: 2026-09-28
Beskrivning
Join our Team
About this opportunity:
Large Language Models produce fluent text that may contain fabricated claims, a failure mode known as hallucination. Estimating when a model is likely to hallucinate is critical for deployment in high-stakes domains.
As LLMs are increasingly integrated into telecommunications infrastructure for network optimisation, fault diagnosis, and autonomous control, ensuring the reliability and trustworthiness of their outputs becomes essential. A model that hallucinates while managing a mobile network could degrade service or mask real faults.
Recent work shows that attention patterns carry a measurable hallucination signal. Attention sinks-tokens that accumulate disproportionate attention from later tokens-correlate with hallucinated outputs. However, important questions remain. Hallucinations can be intrinsic, contradicting provided context, or extrinsic, fabricating absent information. Current attention-based methods handle intrinsic cases well but may struggle with extrinsic cases. The reasons why attention sinks form, and whether they are connected to training and memorisation, also remain unclear.
The goal of this thesis is to evaluate how reliably attention predicts hallucination across tasks and hallucination types, and to develop a risk-estimation framework based on this analysis. The work will use datasets with easily verifiable outputs, such as question-answering benchmarks, mathematical reasoning, and/or code generation.
The thesis will be conducted by one student, corresponds to 30 hp, and is expected to start in early January 2027. The location is Stockholm, Sweden. Supervisor: Konrad Nowosadko.
What you will do:
The skills you bring:
Experience with mechanistic interpretability or previous work with large language models is considered a plus.
Keywords: hallucination detection, attention, large language models, transformers, and mechanistic interpretability.
Why join Ericsson?At Ericsson, you'll have an outstanding opportunity. The chance to use your skills and imagination to push the boundaries of what's possible. To build solutions never seen before to some of the world's toughest problems. You'll be challenged, but you won't be alone. You'll be joining a team of diverse innovators, all driven to go beyond the status quo to craft what comes next.
What happens once you apply? Click Here to find all you need to know about what our typical hiring process looks like.Encouraging a diverse and inclusive organization is core to our values at Ericsson, that's why we champion it in everything we do. We truly believe that by collaborating with people with different experiences we drive innovation, which is essential for our future growth. We encourage people from all backgrounds to apply and realize their full potential as part of our Ericsson team. Ericsson is proud to be an Equal Opportunity Employer. learn more.
Primary country and city: Sweden (SE) || Stockholm
Req ID: 791198
About this opportunity:
Large Language Models produce fluent text that may contain fabricated claims, a failure mode known as hallucination. Estimating when a model is likely to hallucinate is critical for deployment in high-stakes domains.
As LLMs are increasingly integrated into telecommunications infrastructure for network optimisation, fault diagnosis, and autonomous control, ensuring the reliability and trustworthiness of their outputs becomes essential. A model that hallucinates while managing a mobile network could degrade service or mask real faults.
Recent work shows that attention patterns carry a measurable hallucination signal. Attention sinks-tokens that accumulate disproportionate attention from later tokens-correlate with hallucinated outputs. However, important questions remain. Hallucinations can be intrinsic, contradicting provided context, or extrinsic, fabricating absent information. Current attention-based methods handle intrinsic cases well but may struggle with extrinsic cases. The reasons why attention sinks form, and whether they are connected to training and memorisation, also remain unclear.
The goal of this thesis is to evaluate how reliably attention predicts hallucination across tasks and hallucination types, and to develop a risk-estimation framework based on this analysis. The work will use datasets with easily verifiable outputs, such as question-answering benchmarks, mathematical reasoning, and/or code generation.
The thesis will be conducted by one student, corresponds to 30 hp, and is expected to start in early January 2027. The location is Stockholm, Sweden. Supervisor: Konrad Nowosadko.
What you will do:
- Reproduce results from principal attention-based detection methods, such as SinkProbe and Head Entropy.
- Evaluate each method across task types and hallucination types, including intrinsic and extrinsic hallucinations.
- Incorporate the studied methods into a unified framework.
- Develop a hallucination-risk estimation method.
- Analyse the resulting framework by characterising failure modes, performance, and accuracy.
- Optionally, explore why attention sinks form during hallucination, including whether the memorisation status of queried knowledge influences sink formation.
- Deliver a reproducible experimental repository and present the results.
The skills you bring:
- You are a final-year Master's student in Computer Science, Machine Learning, Applied Mathematics, or a related programme.
- You have a solid foundation in machine learning, deep learning, and statistics.
- You have strong Python skills and experience with PyTorch.
- You are familiar with transformer architectures and attention mechanisms.
- You can work independently and communicate your findings clearly.
- You are curious and enthusiastic about exploratory research.
Experience with mechanistic interpretability or previous work with large language models is considered a plus.
Keywords: hallucination detection, attention, large language models, transformers, and mechanistic interpretability.
Why join Ericsson?At Ericsson, you'll have an outstanding opportunity. The chance to use your skills and imagination to push the boundaries of what's possible. To build solutions never seen before to some of the world's toughest problems. You'll be challenged, but you won't be alone. You'll be joining a team of diverse innovators, all driven to go beyond the status quo to craft what comes next.
What happens once you apply? Click Here to find all you need to know about what our typical hiring process looks like.Encouraging a diverse and inclusive organization is core to our values at Ericsson, that's why we champion it in everything we do. We truly believe that by collaborating with people with different experiences we drive innovation, which is essential for our future growth. We encourage people from all backgrounds to apply and realize their full potential as part of our Ericsson team. Ericsson is proud to be an Equal Opportunity Employer. learn more.
Primary country and city: Sweden (SE) || Stockholm
Req ID: 791198
Ansök till tjänsten
Master's Thesis: Attention-Based Hallucination Probability Estimation in Large Language Models
Ny
OM FÖRETAGET
Ericsson AB
