https://megacpp.com 2026-04-19T23:41:31.307Z daily 1 https://megacpp.com/about 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/blog 2026-04-19T23:41:31.307Z daily 0.9 https://megacpp.com/blog/eval 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/contact 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/docs 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/docs/slm 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/people 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/products 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/products/slm 2026-04-19T23:41:31.307Z weekly 0.9 https://megacpp.com/search 2026-04-19T23:41:31.307Z weekly 0.7 https://megacpp.com/blog/ablation-after-10k-steps 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/activation-checkpointing-deep 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/activation-checkpointing-policy 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/activation-recompute-boundaries 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/activations-and-how-we-split-them 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/adapter-system-and-lora-stack 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/attention-sinks-and-telemetry-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/attention-validity-and-structure 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/author-mamba3-spec 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/block-sparse-attention-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/checkpoint-format-and-resume 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/clang-semantic-indexing 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/clustered-sparse-on-tpu-the-planner-stages 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/code-deduplication-at-scale 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/comms-cost-and-overlap 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/compile-commands-and-semantic-graphs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/compile-time-vs-runtime-tradeoffs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/context-parallel-and-sequence-parallel 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/converting-parquet-token-shards-into-megatron-indexed-datasets 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-corpus-construction-detailed 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-data-enhancements-and-why 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-data-preparation-pipeline-deep 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-data-versioning-and-schema 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-eval-suites-and-verifiers 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpp-tokenizer-deep 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cpu-offload-and-memory-calibration-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/cute-dsl-experiments 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/data-pipeline-story 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/data-poisoning-and-refusal-drills 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/data-shuffling-and-seed-discipline 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dataloader-throughput-and-stalls 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dataset-versions-deep-dive 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dataset-versions-v2-to-v6 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/determinism-and-bit-exact-runs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/distillation-bestofn-and-rl 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/distributed-optimizer-stress 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/doc-masking-and-curriculum 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dsa-cuda-graph-safety-deep-dive 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dsa-cuda-graph-safety 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dsa-index-cache-patch 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dsa-indexer-memory-fix-deep-dive 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dsa-indexer-memory-fix 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dualpipe-and-3d-parallelism-on-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/dynamo-and-compile-breakage 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/ep-pp-tp-cp-sp-dp-overview 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/eval-harness-plumbing 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/eval 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/expert-parallel-and-moe-sharding 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/external-library-glitches-we-fixed 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fa4-catalog-on-blackwell 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fail-closed-hybrid-pattern-translation 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fire-dash-redo-in-practice 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fire-plasticity-toolkit 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/flash-attention-4-in-practice 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fp8-and-fp8-rollout 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/framework-survey-fsdp-vs-megatron-vs-deepspeed 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fsdp-cuda-and-megatron-ddp 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fsdp2-on-xla-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fsdp2-pain-and-payoff 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fused-mla-on-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/fused-moe-and-deep-ep-on-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/gated-deltanet-and-hyper-connections 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/gateskip-and-flexidepth-after-the-router 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/gb10-journey 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/gb10-stack-parity-for-megacpp 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/geodesic-loss-design-notes 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/gradient-accumulation-and-microbatching 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/graph-recompilation-hell 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/h200-bringup-and-naming 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/h200-memory-geometry 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/how-to-express-a-nemotron-style-recipe-as-pure-megatron-cli 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/how-we-keep-a-patch-lane 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/hybrid-layer-interleaving 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/inference-serving-stack 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/kernel-catalog-and-impact 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/kernels-that-pay-for-themselves 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/kv-cache-and-paged-attention 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/libtpu-and-jax-interaction 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/libtpu-pjrt-jax-ownership-boundaries 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/license-and-corpus-provenance 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/liger-flce-reduction-none 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/long-context-and-sinks 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/loss-curves-and-divergence-playbook 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/m2rnn-engram-memory 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba-linear-ce-parity-deep-dive 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-fused-trapezoidal-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-hybrid 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-kernel-journey 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-mimo-3d-to-2d-smem-deep-dive 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-parallel-performance 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mamba3-psiv-cache-scaffold 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/manual-splits-and-what-they-cost 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/megacpp-model-glossary 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/megatron-binidx-pipeline 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/megatron-flce-on-hopper 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/memory-budget-anatomy 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mhc-fused-on-blackwell 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/migration-policy-native-megatron-vs-narrow-custom-seams 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mla-and-weight-absorption 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/mod-moda-mtp-stack 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-benchmark-receipts 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-debugging-playbook 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-image-and-cold-start 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-multi-gpu-issues-and-fixes 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-training-platform-overview 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/modal-vs-owned-hardware 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/moe-routing-we-actually-shipped 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/multi-gpu-modal-benchmarks 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/muon-optimizer-on-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/nam56r-launch-policy 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/nam56r-megatron-translation 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/nccl-and-collective-hangs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/nvfp4-inference 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/observability-and-slo-dashboards 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/one-morning-of-bugs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/oom-debugging-playbook-h200 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/oom-on-v6e 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/packed-rows-as-the-real-training-contract 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/pallas-fa-softcap-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/pallas-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/porting-to-megatron-friction 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/precision-recipe-fp16-bf16-fp8-nvfp4 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/profiler-and-receipts 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/profiler-guided-optimization 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/protobuf-2gb-limits-and-streaming-checkpoints 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/public-safe-mla-integration-patterns-for-megatron 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/regional-compile-without-losing-the-plot 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/restoration-without-git-history 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/restoring-a-megatron-training-tree-without-git-history 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/sequence-context-expert-splits 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/shared-mla-adapter-boundaries 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/slm-architecture 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/slm-data 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/slm-training 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/sota-ablation-and-comparison 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/sparse-mla-dimension-generalization 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/sparse-mla-fp8-dispatch 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/specialists 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/speculative-decoding-for-ensembles 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/stp-after-ten-thousand-steps 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/stp-geodesic-regularizer 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/structure-embeddings-and-relation-bias 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/te-replacements-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tensor-parallel-and-sharding 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/throughput-vs-quality-knobs 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tilelang-tma-and-h200-reality 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tilelang-tma-bulk-copy-3d-smem-deep-dive 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tokenized-enriched-pipeline-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tokenizer-v2-v3 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-2-12-journey 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-2-13-on-gb10-serving-and-training-stack 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-212-nightly-wheel-matrix 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-212-xla-breakage-matrix 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-xla-2-11-tpu-reality 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/torch-xla-pjrt-reality 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/tpu-v6e-host-bringup 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/training-on-h200-eight-gpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/training-speed-anatomy-h200 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/training-speed-by-feature 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/transformer-engine-bridge-on-nvidia 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/triton-kernels-we-maintain 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/unique-additions-and-why 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/upstream-pr-mamba3-and-mla 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/upstream-pr-tilelang-and-megatron 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/upstream-prs-overview 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/v6e-performance-deep-dive 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/verifier-first-cpp-evals 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/vllm-gb10-overlay-and-disabled-paths 2026-04-19T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/what-megatron-can-and-cannot-split 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/why-small-model-fills-huge-gpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/xla-adamw-and-flags-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/xla-spmd-sharding-annotations 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/xla-spmd-tokenizer-and-vocab-on-tpu 2026-04-18T00:00:00.000Z weekly 0.8 https://megacpp.com/blog/xla-vs-cuda-stack-decisions 2026-04-18T00:00:00.000Z weekly 0.8