https://submarat.github.io/arena4-capstone-melbo-llama/ 2024-10-05T00:00:00+00:00 https://submarat.github.io/transformers-dont-need-layernorm-at-inference-time/ 2025-07-23T00:00:00+00:00 https://submarat.github.io/replicating-rank-1-emergent-misalignment/ 2026-06-18T00:00:00+00:00 https://submarat.github.io/reproducing-gpt2-small/ 2026-06-25T00:00:00+00:00 https://submarat.github.io/reverse-diffusion-on-a-2d-spiral/ 2026-07-02T00:00:00+00:00 https://submarat.github.io/accelerating-up-qwen-image-edit-2511-by-caching-kv/ 2026-07-09T00:00:00+00:00 https://submarat.github.io/a-toy-assistant-sft-and-dpo/ 2026-07-16T00:00:00+00:00 https://submarat.github.io/how-llms-go-from-base-models-to-assistants/ 2026-07-23T00:00:00+00:00 https://submarat.github.io/scaling-up-a-dit-replicating-peebles-xie/ 2026-07-23T00:00:00+00:00 https://submarat.github.io/about/ https://submarat.github.io/archives/ https://submarat.github.io/ https://submarat.github.io/2018-06-24-Android-Push-Notifications-In-China/ https://submarat.github.io/2019-03-10-contention-free-non-blocking-skip-list/ https://submarat.github.io/2019-03-11-Search-engines/ https://submarat.github.io/2019-03-15-CRUSH-algorithm/