Samuel Albanie

Samuel Albanie

Preparing for AI R&D Automation: Scalable Oversight and Hard Benchmarks.

Descriptions

Self-supervised Vision

Summary: A lecture covering the basics of self-supervision in Computer Vision.
Topics: Self-supervised learning
Date: November, 2023
[slides]

GPT-4

Summary: A description of the GPT-4 technical report by OpenAI.
Topics: GPT-4, Large Language Model, Capabilities, Risks
Date: March, 2023
[slides]

What is the alignment problem?

Summary: A short description of What is the alignment problem? by Jan Leike.
Topics: AI alignment, capabilities, the hard problem of alignment
Date: March, 2023
[slides]

BLOOM: A 176B-Parameter Open-Access Multilingual Language Model

Summary: A description of the the work 'BLOOM: A 176B-Parameter Open-Access Multilingual Language Model' by Le Scao et al. published on arxiv in November 2022 as part of the BigScience Workshop. This work provides an overview of the BLOOM model and the efforts involved in its creation.
Topics: foundation models, large language models, multilingual models
Date: January, 2023
[slides]

B-trees: Samuel’s guide

Summary: Samuel's tutorial on the B-tree data structure.
Topics: B-trees, data structures, algorithms, coding
Date: December, 2022
[slides]

Crosslingual Generalization through Multitask Finetuning (BLOOMZ & mT0)

Summary: A description of the the work 'Crosslingual Generalization through Multitask Finetuning' by Niklas Muennighoff et al. published on arxiv in November 2022 as part of the BigScience Workshop. This work introduced the BLOOMZ and mT0 models.
Topics: multitask finetuning, foundation models, large language models, multilingual models
Date: November, 2022
[slides]

Scaling Instruction-Finetuned Language Models (Flan-PaLM)

Summary: A description of the the work 'Scaling Instruction-Finetuned Language Models' by Hyung Won Chung et al. published on arxiv in October 2022. This work introduced the Flan-PaLM 540B model.
Topics: instruction finetuning, foundation models, large language models
Date: October, 2022
[slides]

ReCo: Retrieve and Co-segment for Zero-Shot Transfer

Summary: A short description of the ReCo framework introduced in the work 'ReCo: Retrieve and Co-segment for Zero-shot Transfer' by G. Shin, W. Xie and S. Albanie, published at NeurIPS in 2022.
Topics: semantic segmentation, foundation models, zero shot transfer
Date: October, 2022
[slides]

Big O notation: Samuel’s guide

Summary: Samuel's tutorial on big O notation and other asymptotic notation (little o, big theta, big omega, little omega).
Topics: Complexity, data structures, algorithms, analysis
Date: October, 2022
[slides]

Note on usage/credits:
Please feel free to use the slides and videos if you find them useful.
If so, please maintain references to the original authors of content where they are noted.