https://docs.waffuru.ai/https://docs.waffuru.ai/butter/https://docs.waffuru.ai/butter/architecture/https://docs.waffuru.ai/butter/benchmarking/https://docs.waffuru.ai/butter/capabilities/https://docs.waffuru.ai/butter/changelog/https://docs.waffuru.ai/butter/chat-templates/https://docs.waffuru.ai/butter/developing/adding-a-model/https://docs.waffuru.ai/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/butter/developing/developing/https://docs.waffuru.ai/butter/developing/nccl/https://docs.waffuru.ai/butter/developing/publishing/https://docs.waffuru.ai/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/butter/developing/rst-testing/https://docs.waffuru.ai/butter/developing/rust-efficiency/https://docs.waffuru.ai/butter/developing/spark-ci/https://docs.waffuru.ai/butter/developing/test-inventory/https://docs.waffuru.ai/butter/developing/testing/https://docs.waffuru.ai/butter/generation-parameters/https://docs.waffuru.ai/butter/installation/https://docs.waffuru.ai/butter/kv-cache/https://docs.waffuru.ai/butter/models/https://docs.waffuru.ai/butter/observability/https://docs.waffuru.ai/butter/performance/https://docs.waffuru.ai/butter/product-packaging/https://docs.waffuru.ai/butter/quantization/https://docs.waffuru.ai/butter/quickstart/https://docs.waffuru.ai/butter/serve/https://docs.waffuru.ai/butter/streaming/https://docs.waffuru.ai/butter/using-the-cli/https://docs.waffuru.ai/es/https://docs.waffuru.ai/es/butter/https://docs.waffuru.ai/es/butter/architecture/https://docs.waffuru.ai/es/butter/benchmarking/https://docs.waffuru.ai/es/butter/capabilities/https://docs.waffuru.ai/es/butter/changelog/https://docs.waffuru.ai/es/butter/chat-templates/https://docs.waffuru.ai/es/butter/developing/adding-a-model/https://docs.waffuru.ai/es/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/es/butter/developing/developing/https://docs.waffuru.ai/es/butter/developing/nccl/https://docs.waffuru.ai/es/butter/developing/publishing/https://docs.waffuru.ai/es/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/es/butter/developing/rst-testing/https://docs.waffuru.ai/es/butter/developing/rust-efficiency/https://docs.waffuru.ai/es/butter/developing/spark-ci/https://docs.waffuru.ai/es/butter/developing/test-inventory/https://docs.waffuru.ai/es/butter/developing/testing/https://docs.waffuru.ai/es/butter/generation-parameters/https://docs.waffuru.ai/es/butter/installation/https://docs.waffuru.ai/es/butter/kv-cache/https://docs.waffuru.ai/es/butter/models/https://docs.waffuru.ai/es/butter/observability/https://docs.waffuru.ai/es/butter/performance/https://docs.waffuru.ai/es/butter/product-packaging/https://docs.waffuru.ai/es/butter/quantization/https://docs.waffuru.ai/es/butter/quickstart/https://docs.waffuru.ai/es/butter/serve/https://docs.waffuru.ai/es/butter/streaming/https://docs.waffuru.ai/es/butter/using-the-cli/https://docs.waffuru.ai/es/iron/https://docs.waffuru.ai/es/iron/benchmarks/https://docs.waffuru.ai/es/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/es/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/es/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/es/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/es/iron/cli/https://docs.waffuru.ai/es/iron/developing/https://docs.waffuru.ai/es/iron/getting-started/https://docs.waffuru.ai/es/iron/kernel-engineering-guide/https://docs.waffuru.ai/es/iron/publishing/https://docs.waffuru.ai/es/iron/q4-projections/https://docs.waffuru.ai/es/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/es/iron/rst-testing/https://docs.waffuru.ai/es/iron/rust-efficiency/https://docs.waffuru.ai/es/iron/spark-ci/https://docs.waffuru.ai/es/iron/specs/amd-backend-spec/https://docs.waffuru.ai/es/iron/specs/ane-backend-spec/https://docs.waffuru.ai/es/iron/specs/architecture/https://docs.waffuru.ai/es/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/es/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/es/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/es/iron/specs/cuda-bench-report/https://docs.waffuru.ai/es/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/es/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/es/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/es/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/es/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/es/iron/specs/kernel-audit/https://docs.waffuru.ai/es/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/es/iron/specs/proposed-optimizations/https://docs.waffuru.ai/es/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/es/iron/specs/tensorops-primitive/https://docs.waffuru.ai/es/iron/specs/toolchain-design/https://docs.waffuru.ai/es/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/es/iron/style-guide/https://docs.waffuru.ai/es/iron/test-inventory/https://docs.waffuru.ai/es/iron/testing/https://docs.waffuru.ai/fr/https://docs.waffuru.ai/fr/butter/https://docs.waffuru.ai/fr/butter/architecture/https://docs.waffuru.ai/fr/butter/benchmarking/https://docs.waffuru.ai/fr/butter/capabilities/https://docs.waffuru.ai/fr/butter/changelog/https://docs.waffuru.ai/fr/butter/chat-templates/https://docs.waffuru.ai/fr/butter/developing/adding-a-model/https://docs.waffuru.ai/fr/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/fr/butter/developing/developing/https://docs.waffuru.ai/fr/butter/developing/nccl/https://docs.waffuru.ai/fr/butter/developing/publishing/https://docs.waffuru.ai/fr/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/fr/butter/developing/rst-testing/https://docs.waffuru.ai/fr/butter/developing/rust-efficiency/https://docs.waffuru.ai/fr/butter/developing/spark-ci/https://docs.waffuru.ai/fr/butter/developing/test-inventory/https://docs.waffuru.ai/fr/butter/developing/testing/https://docs.waffuru.ai/fr/butter/generation-parameters/https://docs.waffuru.ai/fr/butter/installation/https://docs.waffuru.ai/fr/butter/kv-cache/https://docs.waffuru.ai/fr/butter/models/https://docs.waffuru.ai/fr/butter/observability/https://docs.waffuru.ai/fr/butter/performance/https://docs.waffuru.ai/fr/butter/product-packaging/https://docs.waffuru.ai/fr/butter/quantization/https://docs.waffuru.ai/fr/butter/quickstart/https://docs.waffuru.ai/fr/butter/serve/https://docs.waffuru.ai/fr/butter/streaming/https://docs.waffuru.ai/fr/butter/using-the-cli/https://docs.waffuru.ai/fr/iron/https://docs.waffuru.ai/fr/iron/benchmarks/https://docs.waffuru.ai/fr/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/fr/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/fr/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/fr/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/fr/iron/cli/https://docs.waffuru.ai/fr/iron/developing/https://docs.waffuru.ai/fr/iron/getting-started/https://docs.waffuru.ai/fr/iron/kernel-engineering-guide/https://docs.waffuru.ai/fr/iron/publishing/https://docs.waffuru.ai/fr/iron/q4-projections/https://docs.waffuru.ai/fr/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/fr/iron/rst-testing/https://docs.waffuru.ai/fr/iron/rust-efficiency/https://docs.waffuru.ai/fr/iron/spark-ci/https://docs.waffuru.ai/fr/iron/specs/amd-backend-spec/https://docs.waffuru.ai/fr/iron/specs/ane-backend-spec/https://docs.waffuru.ai/fr/iron/specs/architecture/https://docs.waffuru.ai/fr/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/fr/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/fr/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/fr/iron/specs/cuda-bench-report/https://docs.waffuru.ai/fr/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/fr/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/fr/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/fr/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/fr/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/fr/iron/specs/kernel-audit/https://docs.waffuru.ai/fr/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/fr/iron/specs/proposed-optimizations/https://docs.waffuru.ai/fr/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/fr/iron/specs/tensorops-primitive/https://docs.waffuru.ai/fr/iron/specs/toolchain-design/https://docs.waffuru.ai/fr/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/fr/iron/style-guide/https://docs.waffuru.ai/fr/iron/test-inventory/https://docs.waffuru.ai/fr/iron/testing/https://docs.waffuru.ai/iron/https://docs.waffuru.ai/iron/benchmarks/https://docs.waffuru.ai/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/iron/cli/https://docs.waffuru.ai/iron/developing/https://docs.waffuru.ai/iron/getting-started/https://docs.waffuru.ai/iron/kernel-engineering-guide/https://docs.waffuru.ai/iron/publishing/https://docs.waffuru.ai/iron/q4-projections/https://docs.waffuru.ai/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/iron/rst-testing/https://docs.waffuru.ai/iron/rust-efficiency/https://docs.waffuru.ai/iron/spark-ci/https://docs.waffuru.ai/iron/specs/amd-backend-spec/https://docs.waffuru.ai/iron/specs/ane-backend-spec/https://docs.waffuru.ai/iron/specs/architecture/https://docs.waffuru.ai/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/iron/specs/cuda-bench-report/https://docs.waffuru.ai/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/iron/specs/kernel-audit/https://docs.waffuru.ai/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/iron/specs/proposed-optimizations/https://docs.waffuru.ai/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/iron/specs/tensorops-primitive/https://docs.waffuru.ai/iron/specs/toolchain-design/https://docs.waffuru.ai/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/iron/style-guide/https://docs.waffuru.ai/iron/test-inventory/https://docs.waffuru.ai/iron/testing/https://docs.waffuru.ai/it/https://docs.waffuru.ai/it/butter/https://docs.waffuru.ai/it/butter/architecture/https://docs.waffuru.ai/it/butter/benchmarking/https://docs.waffuru.ai/it/butter/capabilities/https://docs.waffuru.ai/it/butter/changelog/https://docs.waffuru.ai/it/butter/chat-templates/https://docs.waffuru.ai/it/butter/developing/adding-a-model/https://docs.waffuru.ai/it/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/it/butter/developing/developing/https://docs.waffuru.ai/it/butter/developing/nccl/https://docs.waffuru.ai/it/butter/developing/publishing/https://docs.waffuru.ai/it/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/it/butter/developing/rst-testing/https://docs.waffuru.ai/it/butter/developing/rust-efficiency/https://docs.waffuru.ai/it/butter/developing/spark-ci/https://docs.waffuru.ai/it/butter/developing/test-inventory/https://docs.waffuru.ai/it/butter/developing/testing/https://docs.waffuru.ai/it/butter/generation-parameters/https://docs.waffuru.ai/it/butter/installation/https://docs.waffuru.ai/it/butter/kv-cache/https://docs.waffuru.ai/it/butter/models/https://docs.waffuru.ai/it/butter/observability/https://docs.waffuru.ai/it/butter/performance/https://docs.waffuru.ai/it/butter/product-packaging/https://docs.waffuru.ai/it/butter/quantization/https://docs.waffuru.ai/it/butter/quickstart/https://docs.waffuru.ai/it/butter/serve/https://docs.waffuru.ai/it/butter/streaming/https://docs.waffuru.ai/it/butter/using-the-cli/https://docs.waffuru.ai/it/iron/https://docs.waffuru.ai/it/iron/benchmarks/https://docs.waffuru.ai/it/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/it/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/it/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/it/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/it/iron/cli/https://docs.waffuru.ai/it/iron/developing/https://docs.waffuru.ai/it/iron/getting-started/https://docs.waffuru.ai/it/iron/kernel-engineering-guide/https://docs.waffuru.ai/it/iron/publishing/https://docs.waffuru.ai/it/iron/q4-projections/https://docs.waffuru.ai/it/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/it/iron/rst-testing/https://docs.waffuru.ai/it/iron/rust-efficiency/https://docs.waffuru.ai/it/iron/spark-ci/https://docs.waffuru.ai/it/iron/specs/amd-backend-spec/https://docs.waffuru.ai/it/iron/specs/ane-backend-spec/https://docs.waffuru.ai/it/iron/specs/architecture/https://docs.waffuru.ai/it/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/it/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/it/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/it/iron/specs/cuda-bench-report/https://docs.waffuru.ai/it/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/it/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/it/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/it/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/it/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/it/iron/specs/kernel-audit/https://docs.waffuru.ai/it/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/it/iron/specs/proposed-optimizations/https://docs.waffuru.ai/it/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/it/iron/specs/tensorops-primitive/https://docs.waffuru.ai/it/iron/specs/toolchain-design/https://docs.waffuru.ai/it/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/it/iron/style-guide/https://docs.waffuru.ai/it/iron/test-inventory/https://docs.waffuru.ai/it/iron/testing/https://docs.waffuru.ai/ja/https://docs.waffuru.ai/ja/butter/https://docs.waffuru.ai/ja/butter/architecture/https://docs.waffuru.ai/ja/butter/benchmarking/https://docs.waffuru.ai/ja/butter/capabilities/https://docs.waffuru.ai/ja/butter/changelog/https://docs.waffuru.ai/ja/butter/chat-templates/https://docs.waffuru.ai/ja/butter/developing/adding-a-model/https://docs.waffuru.ai/ja/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/ja/butter/developing/developing/https://docs.waffuru.ai/ja/butter/developing/nccl/https://docs.waffuru.ai/ja/butter/developing/publishing/https://docs.waffuru.ai/ja/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/ja/butter/developing/rst-testing/https://docs.waffuru.ai/ja/butter/developing/rust-efficiency/https://docs.waffuru.ai/ja/butter/developing/spark-ci/https://docs.waffuru.ai/ja/butter/developing/test-inventory/https://docs.waffuru.ai/ja/butter/developing/testing/https://docs.waffuru.ai/ja/butter/generation-parameters/https://docs.waffuru.ai/ja/butter/installation/https://docs.waffuru.ai/ja/butter/kv-cache/https://docs.waffuru.ai/ja/butter/models/https://docs.waffuru.ai/ja/butter/observability/https://docs.waffuru.ai/ja/butter/performance/https://docs.waffuru.ai/ja/butter/product-packaging/https://docs.waffuru.ai/ja/butter/quantization/https://docs.waffuru.ai/ja/butter/quickstart/https://docs.waffuru.ai/ja/butter/serve/https://docs.waffuru.ai/ja/butter/streaming/https://docs.waffuru.ai/ja/butter/using-the-cli/https://docs.waffuru.ai/ja/iron/https://docs.waffuru.ai/ja/iron/benchmarks/https://docs.waffuru.ai/ja/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/ja/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/ja/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/ja/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/ja/iron/cli/https://docs.waffuru.ai/ja/iron/developing/https://docs.waffuru.ai/ja/iron/getting-started/https://docs.waffuru.ai/ja/iron/kernel-engineering-guide/https://docs.waffuru.ai/ja/iron/publishing/https://docs.waffuru.ai/ja/iron/q4-projections/https://docs.waffuru.ai/ja/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/ja/iron/rst-testing/https://docs.waffuru.ai/ja/iron/rust-efficiency/https://docs.waffuru.ai/ja/iron/spark-ci/https://docs.waffuru.ai/ja/iron/specs/amd-backend-spec/https://docs.waffuru.ai/ja/iron/specs/ane-backend-spec/https://docs.waffuru.ai/ja/iron/specs/architecture/https://docs.waffuru.ai/ja/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/ja/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/ja/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/ja/iron/specs/cuda-bench-report/https://docs.waffuru.ai/ja/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/ja/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/ja/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/ja/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/ja/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/ja/iron/specs/kernel-audit/https://docs.waffuru.ai/ja/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/ja/iron/specs/proposed-optimizations/https://docs.waffuru.ai/ja/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/ja/iron/specs/tensorops-primitive/https://docs.waffuru.ai/ja/iron/specs/toolchain-design/https://docs.waffuru.ai/ja/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/ja/iron/style-guide/https://docs.waffuru.ai/ja/iron/test-inventory/https://docs.waffuru.ai/ja/iron/testing/https://docs.waffuru.ai/ko/https://docs.waffuru.ai/ko/butter/https://docs.waffuru.ai/ko/butter/architecture/https://docs.waffuru.ai/ko/butter/benchmarking/https://docs.waffuru.ai/ko/butter/capabilities/https://docs.waffuru.ai/ko/butter/changelog/https://docs.waffuru.ai/ko/butter/chat-templates/https://docs.waffuru.ai/ko/butter/developing/adding-a-model/https://docs.waffuru.ai/ko/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/ko/butter/developing/developing/https://docs.waffuru.ai/ko/butter/developing/nccl/https://docs.waffuru.ai/ko/butter/developing/publishing/https://docs.waffuru.ai/ko/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/ko/butter/developing/rst-testing/https://docs.waffuru.ai/ko/butter/developing/rust-efficiency/https://docs.waffuru.ai/ko/butter/developing/spark-ci/https://docs.waffuru.ai/ko/butter/developing/test-inventory/https://docs.waffuru.ai/ko/butter/developing/testing/https://docs.waffuru.ai/ko/butter/generation-parameters/https://docs.waffuru.ai/ko/butter/installation/https://docs.waffuru.ai/ko/butter/kv-cache/https://docs.waffuru.ai/ko/butter/models/https://docs.waffuru.ai/ko/butter/observability/https://docs.waffuru.ai/ko/butter/performance/https://docs.waffuru.ai/ko/butter/product-packaging/https://docs.waffuru.ai/ko/butter/quantization/https://docs.waffuru.ai/ko/butter/quickstart/https://docs.waffuru.ai/ko/butter/serve/https://docs.waffuru.ai/ko/butter/streaming/https://docs.waffuru.ai/ko/butter/using-the-cli/https://docs.waffuru.ai/ko/iron/https://docs.waffuru.ai/ko/iron/benchmarks/https://docs.waffuru.ai/ko/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/ko/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/ko/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/ko/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/ko/iron/cli/https://docs.waffuru.ai/ko/iron/developing/https://docs.waffuru.ai/ko/iron/getting-started/https://docs.waffuru.ai/ko/iron/kernel-engineering-guide/https://docs.waffuru.ai/ko/iron/publishing/https://docs.waffuru.ai/ko/iron/q4-projections/https://docs.waffuru.ai/ko/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/ko/iron/rst-testing/https://docs.waffuru.ai/ko/iron/rust-efficiency/https://docs.waffuru.ai/ko/iron/spark-ci/https://docs.waffuru.ai/ko/iron/specs/amd-backend-spec/https://docs.waffuru.ai/ko/iron/specs/ane-backend-spec/https://docs.waffuru.ai/ko/iron/specs/architecture/https://docs.waffuru.ai/ko/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/ko/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/ko/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/ko/iron/specs/cuda-bench-report/https://docs.waffuru.ai/ko/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/ko/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/ko/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/ko/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/ko/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/ko/iron/specs/kernel-audit/https://docs.waffuru.ai/ko/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/ko/iron/specs/proposed-optimizations/https://docs.waffuru.ai/ko/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/ko/iron/specs/tensorops-primitive/https://docs.waffuru.ai/ko/iron/specs/toolchain-design/https://docs.waffuru.ai/ko/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/ko/iron/style-guide/https://docs.waffuru.ai/ko/iron/test-inventory/https://docs.waffuru.ai/ko/iron/testing/https://docs.waffuru.ai/zh/https://docs.waffuru.ai/zh/butter/https://docs.waffuru.ai/zh/butter/architecture/https://docs.waffuru.ai/zh/butter/benchmarking/https://docs.waffuru.ai/zh/butter/capabilities/https://docs.waffuru.ai/zh/butter/changelog/https://docs.waffuru.ai/zh/butter/chat-templates/https://docs.waffuru.ai/zh/butter/developing/adding-a-model/https://docs.waffuru.ai/zh/butter/developing/cuda-selector-campaign/https://docs.waffuru.ai/zh/butter/developing/developing/https://docs.waffuru.ai/zh/butter/developing/nccl/https://docs.waffuru.ai/zh/butter/developing/publishing/https://docs.waffuru.ai/zh/butter/developing/qwen-block-fp8/https://docs.waffuru.ai/zh/butter/developing/rst-testing/https://docs.waffuru.ai/zh/butter/developing/rust-efficiency/https://docs.waffuru.ai/zh/butter/developing/spark-ci/https://docs.waffuru.ai/zh/butter/developing/test-inventory/https://docs.waffuru.ai/zh/butter/developing/testing/https://docs.waffuru.ai/zh/butter/generation-parameters/https://docs.waffuru.ai/zh/butter/installation/https://docs.waffuru.ai/zh/butter/kv-cache/https://docs.waffuru.ai/zh/butter/models/https://docs.waffuru.ai/zh/butter/observability/https://docs.waffuru.ai/zh/butter/performance/https://docs.waffuru.ai/zh/butter/product-packaging/https://docs.waffuru.ai/zh/butter/quantization/https://docs.waffuru.ai/zh/butter/quickstart/https://docs.waffuru.ai/zh/butter/serve/https://docs.waffuru.ai/zh/butter/streaming/https://docs.waffuru.ai/zh/butter/using-the-cli/https://docs.waffuru.ai/zh/iron/https://docs.waffuru.ai/zh/iron/benchmarks/https://docs.waffuru.ai/zh/iron/benchmarks/nax-bt-split-2026-09-06/https://docs.waffuru.ai/zh/iron/benchmarks/nax-ld40-2026-09-06/https://docs.waffuru.ai/zh/iron/benchmarks/nax-wide-2026-09-06/https://docs.waffuru.ai/zh/iron/benchmarks/swiglu-x4-2026-09-05/https://docs.waffuru.ai/zh/iron/cli/https://docs.waffuru.ai/zh/iron/developing/https://docs.waffuru.ai/zh/iron/getting-started/https://docs.waffuru.ai/zh/iron/kernel-engineering-guide/https://docs.waffuru.ai/zh/iron/publishing/https://docs.waffuru.ai/zh/iron/q4-projections/https://docs.waffuru.ai/zh/iron/qwen38-flash-next-kernel-tdd/https://docs.waffuru.ai/zh/iron/rst-testing/https://docs.waffuru.ai/zh/iron/rust-efficiency/https://docs.waffuru.ai/zh/iron/spark-ci/https://docs.waffuru.ai/zh/iron/specs/amd-backend-spec/https://docs.waffuru.ai/zh/iron/specs/ane-backend-spec/https://docs.waffuru.ai/zh/iron/specs/architecture/https://docs.waffuru.ai/zh/iron/specs/bench-metrics-spec/https://docs.waffuru.ai/zh/iron/specs/cuda-backend-scope/https://docs.waffuru.ai/zh/iron/specs/cuda-backend-spec/https://docs.waffuru.ai/zh/iron/specs/cuda-bench-report/https://docs.waffuru.ai/zh/iron/specs/glm53-flash-tdd/https://docs.waffuru.ai/zh/iron/specs/hip-vulkan-phase1/https://docs.waffuru.ai/zh/iron/specs/hip-vulkan-phase2/https://docs.waffuru.ai/zh/iron/specs/hip-vulkan-phase3/https://docs.waffuru.ai/zh/iron/specs/hy3-moe-prep/https://docs.waffuru.ai/zh/iron/specs/kernel-audit/https://docs.waffuru.ai/zh/iron/specs/kernel-consolidation-plan/https://docs.waffuru.ai/zh/iron/specs/proposed-optimizations/https://docs.waffuru.ai/zh/iron/specs/qwen38-flash-next-tdd/https://docs.waffuru.ai/zh/iron/specs/tensorops-primitive/https://docs.waffuru.ai/zh/iron/specs/toolchain-design/https://docs.waffuru.ai/zh/iron/specs/vulkan-backend-spec/https://docs.waffuru.ai/zh/iron/style-guide/https://docs.waffuru.ai/zh/iron/test-inventory/https://docs.waffuru.ai/zh/iron/testing/