September 2024
ML/Stats
- https://cohere.com/blog/intro-prompt-tuner a Cohere tool to iteratively improve prompts
- https://x.com/NeelNanda5/status/1818680642621915527 Gemma Scope announcement
- https://nicholas.carlini.com/writing/2024/how-i-use-ai.html the strongest writing i’ve seen on making llms useful
- https://www.ncei.noaa.gov/access/monitoring/dyk/loess line-smoothing tech
- https://arxiv.org/pdf/2212.14034 high-performance gpu training
- https://statmodeling.stat.columbia.edu/2024/08/21/which-books-papers-and-blogs-are-in-the-bayesian-canon/ starting point for becoming a Bayesian
- https://www.owlposting.com/p/things-i-learned-talking-to-the-new the new non-academic science institutions
- https://writings.stephenwolfram.com/2024/08/whats-really-going-on-in-machine-learning-some-minimal-models wolfram wolframing
- https://www.youtube.com/watch?v=es_8_iT-oQk&list=PLGVZCDnMOq0q7a2aoNP1au_1egfZEjGL6&index=6 sklearn demo from pydata
- https://www.stat.cmu.edu/~cshalizi/TALR/TALR.pdf a close reading of linear regression
- https://hamel.dev/blog/posts/course applied llms
- https://carl-mcbride-ellis.github.io/TOBoML/TOBoML.pdf the orange book of ML
- https://jc.gatspress.com/pdf/otherness_full.pdf ethics and philosophy for AGI
- https://allendowney.github.io/ThinkBayes2/ Bayesian course
- https://linear.axler.net/ Linear Algebra Done Right
- https://arxiv.org/abs//2407.21792 paper on safetywashing
- https://poloclub.github.io/transformer-explainer/ interactive transformer viz
Object Detection:
- https://ai.meta.com/sam2/ meta’s new segmentation model
- https://github.com/facebookresearch/segment-anything/blob/main/notebooks/automatic_mask_generator_example.ipynb tutorials for SAM
- https://github.com/facebookresearch/segment-anything/blob/main/notebooks/onnx_model_example.ipynb as above
- https://github.com/facebookresearch/segment-anything/blob/main/notebooks/predictor_example.ipynb as above
- https://ai.google.dev/gemma/docs/paligemma/fine-tuning-paligemma Google’s multimodal model
- https://arxiv.org/abs/2303.05499 Grounding DINO paper
- https://github.com/NielsRogge/Transformers-Tutorials/blob/master/Grounding DINO/Inference_with_Grounding_DINO_for_zero_shot_object_detection.ipynb Grounding DINO example
- https://huggingface.co/docs/transformers/model_doc/grounding-dino hugging face grounding DINO
- https://pytorch.org/vision/stable/auto_examples/others/plot_repurposing_annotations.html pytorch vision demo
- https://pytorch.org/vision/stable/auto_examples/transforms/plot_transforms_e2e.html pytorch vision demo 2
- https://pytorch.org/vision/stable/models.html#semantic-segmentation pytorch vision demo 3
- https://github.com/simonw/tools/blob/main/gemini-bbox.html using Gemini for bounding boxes in the browser.
Bluedot Readings:
- https://lilianweng.github.io/posts/2018-02-19-rl-overview/ reinforcement learning overview
- http://colah.github.io/ really good basic ML blogs
- http://colah.github.io/posts/2014-03-NN-Manifolds-Topology/ proper maths on neural networks
- https://distill.pub/2017/momentum/ explanation of momentum in gradient descent
- https://www.3blue1brown.com/topics/neural-networks explainer videos
- https://www.deep-ml.com code challenges for testing ML knowledge
- https://devinterp.com/projects example starter projects in dev interp
- https://www.youtube.com/watch?v=R3nbXgMnVqQ walkthough of superposition paper
- https://www.alignmentforum.org/posts/LNA8mubrByG7SFacm/against-almost-every-theory-of-impact-of-interpretability-1 good mech. interp. critique
- https://storage.googleapis.com/jumprelu-saes-paper/JumpReLU_SAEs.pdf improvements on SAEs
- https://www.youtube.com/watch?v=bZvPLRZx-V8 more mathsy mech. interp. video
Misc
- https://paulgraham.com/superlinear.html PG exhortations
- https://paulgraham.com/persistence.html PG further exhortations
- https://paulgraham.com/greatwork.html PG even further exhortations
- https://www.youtube.com/watch?v=Vm49oIPtyRE Dwarkesh on his habits
- https://www.henrikkarlsson.xyz/p/search-query an explanation of blogging
- https://www.henrikkarlsson.xyz/p/multi-armed-bandit an explanation of prioritisation
- https://www.henrikkarlsson.xyz/p/good-ideas an explanation of the idea zone
- https://www.henrikkarlsson.xyz/p/writing-to-think another explanation of blogging.
- a ton of Nanda blog posts