https://richardcheam.github.io/blog/foundations-neurons-backprop/2026-10-15https://richardcheam.github.io/blog/grace-hopper-file-cache/2026-10-03https://richardcheam.github.io/blog/model-development-one-update/2026-11-09https://richardcheam.github.io/blog/benchmark-denominators/2026-10-03https://richardcheam.github.io/blog/foundations-loss-gradient-descent/2026-10-15https://richardcheam.github.io/blog/model-development-pretraining-corpus/2026-11-10https://richardcheam.github.io/blog/foundations-optimisers/2026-10-15https://richardcheam.github.io/blog/gh200-locality-evidence/2026-10-03https://richardcheam.github.io/blog/model-development-data-mixtures/2026-11-11https://richardcheam.github.io/blog/foundations-regularisation/2026-10-15https://richardcheam.github.io/blog/kv-offload-correctness/2026-10-03https://richardcheam.github.io/blog/model-development-tokenization/2026-11-11https://richardcheam.github.io/blog/foundations-normalisation/2026-10-15https://richardcheam.github.io/blog/model-development-decoder-architecture/2026-11-11https://richardcheam.github.io/blog/startup-artifacts-runtime-contract/2026-10-03https://richardcheam.github.io/blog/foundations-embeddings/2026-10-15https://richardcheam.github.io/blog/model-development-small-training-run/2026-11-11https://richardcheam.github.io/blog/throughput-versus-usable-latency/2026-10-03https://richardcheam.github.io/blog/foundations-convolutions/2026-10-15https://richardcheam.github.io/blog/inference-workload-lifecycle/2026-10-03https://richardcheam.github.io/blog/model-development-scaling-laws/2026-11-11https://richardcheam.github.io/blog/foundations-recurrent-networks/2026-10-15https://richardcheam.github.io/blog/model-development-gpu-memory-compute/2026-11-11https://richardcheam.github.io/blog/representation-collapse-explained/2026-10-07https://richardcheam.github.io/blog/audio-basics/2026-10-14https://richardcheam.github.io/blog/residual-vector-quantization/2026-10-14https://richardcheam.github.io/blog/foundations-lstm-gru/2026-10-15https://richardcheam.github.io/blog/model-development-data-parallelism/2026-11-11https://richardcheam.github.io/blog/pocket-tts-calm-explained/2026-10-11https://richardcheam.github.io/blog/foundations-attention/2026-10-15https://richardcheam.github.io/blog/mimi-codec-explained/2026-10-11https://richardcheam.github.io/blog/model-development-sharding-recomputation/2026-11-11https://richardcheam.github.io/blog/cascade-vs-full-duplex/2026-10-12https://richardcheam.github.io/blog/foundations-transformer/2026-10-15https://richardcheam.github.io/blog/model-development-model-parallelism/2026-11-11https://richardcheam.github.io/blog/foundations-autoencoders-vae/2026-10-15https://richardcheam.github.io/blog/model-development-efficient-execution/2026-11-11https://richardcheam.github.io/blog/speech-evaluation-metrics/2026-10-12https://richardcheam.github.io/blog/khmer-script-for-tts/2026-10-14https://richardcheam.github.io/blog/foundations-gans/2026-10-15https://richardcheam.github.io/blog/khmer-tts-fine-tuning/2026-10-13https://richardcheam.github.io/blog/model-development-running-at-scale/2026-11-11https://richardcheam.github.io/blog/foundations-diffusion-flow/2026-10-15https://richardcheam.github.io/blog/model-development-supervised-fine-tuning/2026-11-11https://richardcheam.github.io/blog/foundations-language-models/2026-10-15https://richardcheam.github.io/blog/model-development-post-training-data/2026-11-11https://richardcheam.github.io/blog/model-development-preference-optimization/2026-11-11https://richardcheam.github.io/blog/model-development-reinforcement-learning/2026-11-11https://richardcheam.github.io/blog/model-development-post-training-failures/2026-11-11https://richardcheam.github.io/blog/model-development-choosing-a-model/2026-11-11https://richardcheam.github.io/blog/model-development-prompting-context/2026-11-11https://richardcheam.github.io/blog/model-development-retrieval-tools/2026-11-11https://richardcheam.github.io/blog/model-development-adapt-or-prompt/2026-11-11https://richardcheam.github.io/blog/model-development-inference-serving/2026-11-11https://richardcheam.github.io/blog/model-development-evaluating-applications/2026-11-11https://richardcheam.github.io/blog/model-development-how-it-fits/2026-11-11https://richardcheam.github.io/blog/model-development-other-modalities/2026-11-11https://richardcheam.github.io/paper-reviews/chinchilla/2026-11-11https://richardcheam.github.io/paper-reviews/deepseekmath-grpo/2026-11-11https://richardcheam.github.io/paper-reviews/dpo/2026-11-11https://richardcheam.github.io/paper-reviews/flashattention/2026-11-11https://richardcheam.github.io/paper-reviews/lora/2026-11-11https://richardcheam.github.io/paper-reviews/tulu-3/2026-11-11https://richardcheam.github.io/paper-reviews/fineweb/2026-11-10https://richardcheam.github.io/paper-reviews/megatron-ptd-p/2026-11-09https://richardcheam.github.io/paper-reviews/zero/2026-11-09https://richardcheam.github.io/paper-reviews/dsm/2026-10-14https://richardcheam.github.io/paper-reviews/f5-tts/2026-10-14https://richardcheam.github.io/paper-reviews/flow-matching/2026-10-14https://richardcheam.github.io/paper-reviews/moshi/2026-10-14https://richardcheam.github.io/paper-reviews/soundstream/2026-10-14https://richardcheam.github.io/paper-reviews/vall-e/2026-10-14https://richardcheam.github.io/paper-reviews/whisper/2026-10-14https://richardcheam.github.io/paper-reviews/moshirag/2026-10-12https://richardcheam.github.io/https://richardcheam.github.io/blog/https://richardcheam.github.io/contact/https://richardcheam.github.io/cv/https://richardcheam.github.io/paper-reviews/https://richardcheam.github.io/projects/https://richardcheam.github.io/projects/decisionops/https://richardcheam.github.io/projects/embedding-diagnostics/https://richardcheam.github.io/projects/enigmatica/https://richardcheam.github.io/projects/imdb-moe/https://richardcheam.github.io/projects/khmer-tts/https://richardcheam.github.io/projects/mnist-ssl/https://richardcheam.github.io/projects/xai-vision-transfer/