cui-llama.rn 1.6.0 → 1.7.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +35 -7
- package/android/src/main/CMakeLists.txt +22 -11
- package/android/src/main/java/com/rnllama/LlamaContext.java +42 -6
- package/android/src/main/java/com/rnllama/RNLlama.java +139 -4
- package/android/src/main/jni.cpp +173 -18
- package/android/src/main/jniLibs/arm64-v8a/librnllama.so +0 -0
- package/android/src/main/jniLibs/arm64-v8a/librnllama_v8.so +0 -0
- package/android/src/main/jniLibs/arm64-v8a/librnllama_v8_2.so +0 -0
- package/android/src/main/jniLibs/arm64-v8a/librnllama_v8_2_dotprod.so +0 -0
- package/android/src/main/jniLibs/arm64-v8a/librnllama_v8_2_dotprod_i8mm.so +0 -0
- package/android/src/main/jniLibs/arm64-v8a/librnllama_v8_2_i8mm.so +0 -0
- package/android/src/main/jniLibs/x86_64/librnllama.so +0 -0
- package/android/src/main/jniLibs/x86_64/librnllama_x86_64.so +0 -0
- package/android/src/newarch/java/com/rnllama/RNLlamaModule.java +24 -4
- package/android/src/oldarch/java/com/rnllama/RNLlamaModule.java +22 -2
- package/cpp/LICENSE +21 -0
- package/cpp/chat.cpp +129 -107
- package/cpp/chat.h +2 -0
- package/cpp/common.cpp +58 -78
- package/cpp/common.h +29 -21
- package/cpp/ggml-alloc.c +4 -1
- package/cpp/ggml-backend.cpp +9 -5
- package/cpp/ggml-backend.h +4 -4
- package/cpp/ggml-cpp.h +1 -1
- package/cpp/ggml-cpu/amx/amx.cpp +221 -0
- package/cpp/ggml-cpu/amx/amx.h +8 -0
- package/cpp/ggml-cpu/amx/common.h +91 -0
- package/cpp/ggml-cpu/amx/mmq.cpp +2511 -0
- package/cpp/ggml-cpu/amx/mmq.h +10 -0
- package/{ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers → cpp/ggml-cpu}/binary-ops.h +1 -1
- package/cpp/ggml-cpu/common.h +72 -0
- package/cpp/{ggml-cpu-aarch64.cpp → ggml-cpu/ggml-cpu-aarch64.cpp} +809 -103
- package/cpp/{ggml-cpu-quants.c → ggml-cpu/ggml-cpu-quants.c} +306 -6
- package/cpp/{ggml-cpu.c → ggml-cpu/ggml-cpu.c} +114 -55
- package/cpp/{ggml-cpu.cpp → ggml-cpu/ggml-cpu.cpp} +32 -16
- package/cpp/{ops.cpp → ggml-cpu/ops.cpp} +353 -173
- package/{ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers → cpp/ggml-cpu}/ops.h +2 -20
- package/cpp/{sgemm.cpp → ggml-cpu/sgemm.cpp} +501 -0
- package/{ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers → cpp/ggml-cpu}/simd-mappings.h +7 -3
- package/{ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers → cpp/ggml-cpu}/unary-ops.h +1 -1
- package/cpp/{vec.cpp → ggml-cpu/vec.cpp} +0 -6
- package/{ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers → cpp/ggml-cpu}/vec.h +16 -0
- package/cpp/ggml-cpu.h +5 -0
- package/cpp/ggml-impl.h +16 -9
- package/cpp/ggml-llama-sim.metallib +0 -0
- package/cpp/ggml-llama.metallib +0 -0
- package/cpp/ggml-metal-impl.h +36 -11
- package/cpp/ggml-metal.m +810 -176
- package/cpp/ggml-opt.cpp +373 -190
- package/cpp/ggml-opt.h +49 -28
- package/cpp/ggml-quants.c +0 -6
- package/cpp/ggml.c +227 -282
- package/cpp/ggml.h +82 -101
- package/cpp/gguf.cpp +33 -33
- package/cpp/json-schema-to-grammar.cpp +3 -0
- package/cpp/llama-adapter.cpp +6 -0
- package/cpp/llama-arch.cpp +49 -17
- package/cpp/llama-arch.h +9 -0
- package/cpp/llama-batch.cpp +8 -2
- package/cpp/llama-batch.h +2 -1
- package/cpp/llama-chat.cpp +39 -16
- package/cpp/llama-chat.h +4 -2
- package/cpp/llama-context.cpp +440 -611
- package/cpp/llama-context.h +44 -33
- package/cpp/llama-cparams.h +1 -0
- package/cpp/llama-graph.cpp +214 -291
- package/cpp/llama-graph.h +69 -21
- package/cpp/llama-hparams.cpp +17 -1
- package/cpp/llama-hparams.h +39 -5
- package/cpp/llama-kv-cache.cpp +2067 -620
- package/cpp/llama-kv-cache.h +410 -108
- package/cpp/llama-memory.h +12 -1
- package/cpp/llama-model-loader.cpp +24 -15
- package/cpp/llama-model-saver.cpp +281 -0
- package/cpp/llama-model-saver.h +37 -0
- package/cpp/llama-model.cpp +1089 -359
- package/cpp/llama-model.h +19 -3
- package/cpp/llama-sampling.cpp +20 -7
- package/cpp/llama-vocab.cpp +54 -9
- package/cpp/llama-vocab.h +6 -0
- package/cpp/llama.cpp +14 -0
- package/cpp/llama.h +86 -142
- package/cpp/minja/chat-template.hpp +9 -5
- package/cpp/minja/minja.hpp +69 -36
- package/cpp/rn-llama.cpp +602 -190
- package/cpp/rn-llama.h +34 -8
- package/cpp/sampling.cpp +57 -50
- package/cpp/tools/mtmd/clip-impl.h +462 -0
- package/cpp/tools/mtmd/clip.cpp +4024 -0
- package/cpp/tools/mtmd/clip.h +101 -0
- package/cpp/tools/mtmd/miniaudio.h +93468 -0
- package/cpp/tools/mtmd/mtmd-audio.cpp +855 -0
- package/cpp/tools/mtmd/mtmd-audio.h +62 -0
- package/cpp/tools/mtmd/mtmd-helper.cpp +297 -0
- package/cpp/tools/mtmd/mtmd.cpp +942 -0
- package/cpp/tools/mtmd/mtmd.h +362 -0
- package/cpp/tools/mtmd/stb_image.h +7988 -0
- package/ios/CMakeLists.txt +20 -10
- package/ios/RNLlama.h +6 -0
- package/ios/RNLlama.mm +82 -3
- package/ios/RNLlamaContext.h +5 -1
- package/ios/RNLlamaContext.mm +131 -38
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/chat.h +2 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/common.h +29 -21
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-backend.h +4 -4
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpp.h +1 -1
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpu.h +5 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-impl.h +16 -9
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-metal-impl.h +36 -11
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-opt.h +49 -28
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml.h +82 -101
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-arch.h +9 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-batch.h +2 -1
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-chat.h +4 -2
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-context.h +44 -33
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-cparams.h +1 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-graph.h +69 -21
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-hparams.h +39 -5
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-kv-cache.h +410 -108
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-memory.h +12 -1
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-model-saver.h +37 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-model.h +19 -3
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama-vocab.h +6 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/llama.h +86 -142
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/minja/chat-template.hpp +9 -5
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/minja/minja.hpp +69 -36
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/rn-llama.h +34 -8
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Info.plist +0 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/ggml-llama.metallib +0 -0
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/rnllama +0 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/chat.h +2 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/common.h +29 -21
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-backend.h +4 -4
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpp.h +1 -1
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu.h +5 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-impl.h +16 -9
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-metal-impl.h +36 -11
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-opt.h +49 -28
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml.h +82 -101
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-arch.h +9 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-batch.h +2 -1
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-chat.h +4 -2
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-context.h +44 -33
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-cparams.h +1 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-graph.h +69 -21
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-hparams.h +39 -5
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-kv-cache.h +410 -108
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-memory.h +12 -1
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-model-saver.h +37 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-model.h +19 -3
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama-vocab.h +6 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/llama.h +86 -142
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/minja/chat-template.hpp +9 -5
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/minja/minja.hpp +69 -36
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/rn-llama.h +34 -8
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Info.plist +0 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/_CodeSignature/CodeResources +1 -1
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/ggml-llama-sim.metallib +0 -0
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/rnllama +0 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/chat.h +2 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/common.h +29 -21
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-backend.h +4 -4
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpp.h +1 -1
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpu.h +5 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-impl.h +16 -9
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-metal-impl.h +36 -11
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-opt.h +49 -28
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml.h +82 -101
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-arch.h +9 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-batch.h +2 -1
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-chat.h +4 -2
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-context.h +44 -33
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-cparams.h +1 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-graph.h +69 -21
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-hparams.h +39 -5
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-kv-cache.h +410 -108
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-memory.h +12 -1
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-model-saver.h +37 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-model.h +19 -3
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama-vocab.h +6 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/llama.h +86 -142
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/minja/chat-template.hpp +9 -5
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/minja/minja.hpp +69 -36
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/rn-llama.h +34 -8
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Info.plist +0 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/ggml-llama.metallib +0 -0
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/rnllama +0 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/chat.h +2 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/common.h +29 -21
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-backend.h +4 -4
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpp.h +1 -1
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu.h +5 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-impl.h +16 -9
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-metal-impl.h +36 -11
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-opt.h +49 -28
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml.h +82 -101
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-arch.h +9 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-batch.h +2 -1
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-chat.h +4 -2
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-context.h +44 -33
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-cparams.h +1 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-graph.h +69 -21
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-hparams.h +39 -5
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-kv-cache.h +410 -108
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-memory.h +12 -1
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-model-saver.h +37 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-model.h +19 -3
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama-vocab.h +6 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/llama.h +86 -142
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/minja/chat-template.hpp +9 -5
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/minja/minja.hpp +69 -36
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/rn-llama.h +34 -8
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Info.plist +0 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/_CodeSignature/CodeResources +1 -1
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/ggml-llama-sim.metallib +0 -0
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/rnllama +0 -0
- package/jest/mock.js +33 -7
- package/lib/commonjs/NativeRNLlama.js.map +1 -1
- package/lib/commonjs/index.js +153 -21
- package/lib/commonjs/index.js.map +1 -1
- package/lib/module/NativeRNLlama.js.map +1 -1
- package/lib/module/index.js +152 -20
- package/lib/module/index.js.map +1 -1
- package/lib/typescript/NativeRNLlama.d.ts +54 -4
- package/lib/typescript/NativeRNLlama.d.ts.map +1 -1
- package/lib/typescript/index.d.ts +72 -6
- package/lib/typescript/index.d.ts.map +1 -1
- package/package.json +1 -1
- package/src/NativeRNLlama.ts +72 -4
- package/src/index.ts +212 -38
- package/cpp/binary-ops.h +0 -16
- package/cpp/ops.h +0 -128
- package/cpp/simd-mappings.h +0 -888
- package/cpp/unary-ops.h +0 -28
- package/cpp/vec.h +0 -802
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/binary-ops.h +0 -16
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpu-aarch64.h +0 -8
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpu-impl.h +0 -512
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpu-quants.h +0 -63
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ggml-cpu-traits.h +0 -38
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/ops.h +0 -128
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/sgemm.h +0 -14
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/simd-mappings.h +0 -888
- package/ios/rnllama.xcframework/ios-arm64/rnllama.framework/Headers/vec.h +0 -802
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-aarch64.h +0 -8
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-impl.h +0 -512
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-quants.h +0 -63
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-traits.h +0 -38
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/sgemm.h +0 -14
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/unary-ops.h +0 -28
- package/ios/rnllama.xcframework/ios-arm64_x86_64-simulator/rnllama.framework/Headers/vec.h +0 -802
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/binary-ops.h +0 -16
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpu-aarch64.h +0 -8
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpu-impl.h +0 -512
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpu-quants.h +0 -63
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ggml-cpu-traits.h +0 -38
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/ops.h +0 -128
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/sgemm.h +0 -14
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/simd-mappings.h +0 -888
- package/ios/rnllama.xcframework/tvos-arm64/rnllama.framework/Headers/unary-ops.h +0 -28
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/binary-ops.h +0 -16
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-aarch64.h +0 -8
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-impl.h +0 -512
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-quants.h +0 -63
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ggml-cpu-traits.h +0 -38
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/ops.h +0 -128
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/sgemm.h +0 -14
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/simd-mappings.h +0 -888
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/unary-ops.h +0 -28
- package/ios/rnllama.xcframework/tvos-arm64_x86_64-simulator/rnllama.framework/Headers/vec.h +0 -802
- package/lib/commonjs/chat.js +0 -37
- package/lib/commonjs/chat.js.map +0 -1
- package/lib/module/chat.js +0 -33
- package/lib/module/chat.js.map +0 -1
- package/lib/typescript/chat.d.ts +0 -10
- package/lib/typescript/chat.d.ts.map +0 -1
- package/src/chat.ts +0 -44
- /package/cpp/{binary-ops.cpp → ggml-cpu/binary-ops.cpp} +0 -0
- /package/cpp/{ggml-cpu-aarch64.h → ggml-cpu/ggml-cpu-aarch64.h} +0 -0
- /package/cpp/{ggml-cpu-impl.h → ggml-cpu/ggml-cpu-impl.h} +0 -0
- /package/cpp/{ggml-cpu-quants.h → ggml-cpu/ggml-cpu-quants.h} +0 -0
- /package/cpp/{ggml-cpu-traits.cpp → ggml-cpu/ggml-cpu-traits.cpp} +0 -0
- /package/cpp/{ggml-cpu-traits.h → ggml-cpu/ggml-cpu-traits.h} +0 -0
- /package/cpp/{sgemm.h → ggml-cpu/sgemm.h} +0 -0
- /package/cpp/{unary-ops.cpp → ggml-cpu/unary-ops.cpp} +0 -0
package/cpp/ggml-metal-impl.h
CHANGED
@@ -207,6 +207,10 @@ typedef struct {
|
|
207
207
|
float attn_factor;
|
208
208
|
float beta_fast;
|
209
209
|
float beta_slow;
|
210
|
+
int32_t sect_0;
|
211
|
+
int32_t sect_1;
|
212
|
+
int32_t sect_2;
|
213
|
+
int32_t sect_3;
|
210
214
|
} lm_ggml_metal_kargs_rope;
|
211
215
|
|
212
216
|
typedef struct {
|
@@ -299,21 +303,42 @@ typedef struct {
|
|
299
303
|
} lm_ggml_metal_kargs_mul_mv_ext;
|
300
304
|
|
301
305
|
typedef struct {
|
302
|
-
int32_t
|
303
|
-
int32_t
|
304
|
-
uint64_t
|
306
|
+
int32_t ne10;
|
307
|
+
int32_t ne11; // n_expert_used (bcast)
|
308
|
+
uint64_t nb11;
|
309
|
+
uint64_t nb12;
|
310
|
+
int32_t neh11; // n_tokens
|
311
|
+
uint64_t nbh11;
|
312
|
+
int32_t ne20; // n_expert_used
|
313
|
+
uint64_t nb21;
|
314
|
+
} lm_ggml_metal_kargs_mul_mm_id_map0;
|
315
|
+
|
316
|
+
typedef struct {
|
317
|
+
int32_t ne20; // n_expert_used
|
318
|
+
int32_t neh0;
|
319
|
+
int32_t neh1;
|
320
|
+
uint64_t nbh1;
|
321
|
+
uint64_t nbh2;
|
322
|
+
int32_t ne0;
|
323
|
+
uint64_t nb1;
|
324
|
+
uint64_t nb2;
|
325
|
+
} lm_ggml_metal_kargs_mul_mm_id_map1;
|
326
|
+
|
327
|
+
typedef struct {
|
305
328
|
int32_t ne00;
|
306
329
|
int32_t ne02;
|
307
330
|
uint64_t nb01;
|
308
331
|
uint64_t nb02;
|
309
|
-
|
310
|
-
int32_t
|
311
|
-
|
312
|
-
uint64_t
|
313
|
-
uint64_t
|
314
|
-
uint64_t
|
315
|
-
int32_t
|
316
|
-
int32_t
|
332
|
+
uint64_t nb03;
|
333
|
+
int32_t neh12;
|
334
|
+
uint64_t nbh10;
|
335
|
+
uint64_t nbh11;
|
336
|
+
uint64_t nbh12;
|
337
|
+
uint64_t nbh13;
|
338
|
+
int32_t neh0;
|
339
|
+
int32_t neh1;
|
340
|
+
int16_t r2;
|
341
|
+
int16_t r3;
|
317
342
|
} lm_ggml_metal_kargs_mul_mm_id;
|
318
343
|
|
319
344
|
typedef struct {
|