@novastera-oss/llamarn 0.2.9 → 0.3.0

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.
Files changed (247) hide show
  1. package/android/src/main/jniLibs/arm64-v8a/libggml-base.so +0 -0
  2. package/android/src/main/jniLibs/arm64-v8a/libggml-cpu.so +0 -0
  3. package/android/src/main/jniLibs/arm64-v8a/libggml.so +0 -0
  4. package/android/src/main/jniLibs/arm64-v8a/libllama.so +0 -0
  5. package/android/src/main/jniLibs/armeabi-v7a/libggml-base.so +0 -0
  6. package/android/src/main/jniLibs/armeabi-v7a/libggml-cpu.so +0 -0
  7. package/android/src/main/jniLibs/armeabi-v7a/libggml.so +0 -0
  8. package/android/src/main/jniLibs/armeabi-v7a/libllama.so +0 -0
  9. package/android/src/main/jniLibs/x86/libggml-base.so +0 -0
  10. package/android/src/main/jniLibs/x86/libggml-cpu.so +0 -0
  11. package/android/src/main/jniLibs/x86/libggml.so +0 -0
  12. package/android/src/main/jniLibs/x86/libllama.so +0 -0
  13. package/android/src/main/jniLibs/x86_64/libggml-base.so +0 -0
  14. package/android/src/main/jniLibs/x86_64/libggml-cpu.so +0 -0
  15. package/android/src/main/jniLibs/x86_64/libggml.so +0 -0
  16. package/android/src/main/jniLibs/x86_64/libllama.so +0 -0
  17. package/cpp/build-info.cpp +2 -2
  18. package/cpp/llama.cpp/CMakeLists.txt +0 -1
  19. package/cpp/llama.cpp/README.md +4 -5
  20. package/cpp/llama.cpp/build-xcframework.sh +1 -1
  21. package/cpp/llama.cpp/common/CMakeLists.txt +4 -5
  22. package/cpp/llama.cpp/common/arg.cpp +17 -0
  23. package/cpp/llama.cpp/common/chat.cpp +37 -20
  24. package/cpp/llama.cpp/common/chat.h +2 -0
  25. package/cpp/llama.cpp/common/common.h +4 -0
  26. package/cpp/llama.cpp/convert_hf_to_gguf.py +745 -6
  27. package/cpp/llama.cpp/convert_hf_to_gguf_update.py +9 -0
  28. package/cpp/llama.cpp/ggml/CMakeLists.txt +7 -2
  29. package/cpp/llama.cpp/ggml/include/ggml-backend.h +1 -1
  30. package/cpp/llama.cpp/ggml/include/ggml.h +173 -10
  31. package/cpp/llama.cpp/ggml/src/CMakeLists.txt +0 -1
  32. package/cpp/llama.cpp/ggml/src/ggml-backend-reg.cpp +0 -8
  33. package/cpp/llama.cpp/ggml/src/ggml-backend.cpp +36 -18
  34. package/cpp/llama.cpp/ggml/src/ggml-cann/aclnn_ops.cpp +68 -5
  35. package/cpp/llama.cpp/ggml/src/ggml-cann/ggml-cann.cpp +16 -2
  36. package/cpp/llama.cpp/ggml/src/ggml-cpu/CMakeLists.txt +6 -1
  37. package/cpp/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c +28 -1
  38. package/cpp/llama.cpp/ggml/src/ggml-cpu/ops.cpp +1203 -163
  39. package/cpp/llama.cpp/ggml/src/ggml-cpu/ops.h +6 -0
  40. package/cpp/llama.cpp/ggml/src/ggml-cpu/simd-mappings.h +1 -1
  41. package/cpp/llama.cpp/ggml/src/ggml-cpu/vec.cpp +33 -9
  42. package/cpp/llama.cpp/ggml/src/ggml-cpu/vec.h +142 -9
  43. package/cpp/llama.cpp/ggml/src/ggml-cuda/common.cuh +17 -0
  44. package/cpp/llama.cpp/ggml/src/ggml-cuda/convert.cu +22 -0
  45. package/cpp/llama.cpp/ggml/src/ggml-cuda/convert.cuh +5 -0
  46. package/cpp/llama.cpp/ggml/src/ggml-cuda/cross-entropy-loss.cu +2 -14
  47. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-common.cuh +4 -1
  48. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-mma-f16.cuh +8 -4
  49. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-tile-f16.cu +6 -4
  50. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-tile-f32.cu +14 -12
  51. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-vec-f16.cuh +5 -3
  52. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-vec-f32.cuh +15 -10
  53. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-wmma-f16.cu +8 -6
  54. package/cpp/llama.cpp/ggml/src/ggml-cuda/getrows.cu +8 -0
  55. package/cpp/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu +185 -79
  56. package/cpp/llama.cpp/ggml/src/ggml-cuda/mmq.cuh +2 -8
  57. package/cpp/llama.cpp/ggml/src/ggml-cuda/rope.cu +21 -27
  58. package/cpp/llama.cpp/ggml/src/ggml-cuda/scale.cu +8 -6
  59. package/cpp/llama.cpp/ggml/src/ggml-cuda/softmax.cu +119 -58
  60. package/cpp/llama.cpp/ggml/src/ggml-cuda/ssm-conv.cu +10 -2
  61. package/cpp/llama.cpp/ggml/src/ggml-cuda/ssm-scan.cu +192 -52
  62. package/cpp/llama.cpp/ggml/src/ggml-cuda/unary.cu +97 -0
  63. package/cpp/llama.cpp/ggml/src/ggml-cuda/unary.cuh +11 -0
  64. package/cpp/llama.cpp/ggml/src/ggml-cuda/upscale.cu +92 -6
  65. package/cpp/llama.cpp/ggml/src/ggml-cuda/vendors/hip.h +14 -5
  66. package/cpp/llama.cpp/ggml/src/ggml-impl.h +64 -0
  67. package/cpp/llama.cpp/ggml/src/ggml-metal/CMakeLists.txt +4 -2
  68. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal-impl.h +35 -9
  69. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal.m +167 -39
  70. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal.metal +254 -57
  71. package/cpp/llama.cpp/ggml/src/ggml-opencl/CMakeLists.txt +3 -0
  72. package/cpp/llama.cpp/ggml/src/ggml-opencl/ggml-opencl.cpp +505 -40
  73. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/gelu.cl +27 -0
  74. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/glu.cl +337 -0
  75. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/mul_mat_f16_f32.cl +130 -0
  76. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/scale.cl +3 -2
  77. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/set_rows.cl +95 -0
  78. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/softmax_4_f16.cl +24 -11
  79. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/softmax_4_f32.cl +24 -11
  80. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/softmax_f16.cl +24 -11
  81. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/softmax_f32.cl +24 -11
  82. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/upscale.cl +2 -3
  83. package/cpp/llama.cpp/ggml/src/ggml-quants.c +6 -6
  84. package/cpp/llama.cpp/ggml/src/ggml-sycl/backend.hpp +1 -0
  85. package/cpp/llama.cpp/ggml/src/ggml-sycl/element_wise.cpp +693 -1034
  86. package/cpp/llama.cpp/ggml/src/ggml-sycl/element_wise.hpp +18 -9
  87. package/cpp/llama.cpp/ggml/src/ggml-sycl/ggml-sycl.cpp +60 -9
  88. package/cpp/llama.cpp/ggml/src/ggml-sycl/rope.cpp +15 -18
  89. package/cpp/llama.cpp/ggml/src/ggml-sycl/set_rows.cpp +131 -0
  90. package/cpp/llama.cpp/ggml/src/ggml-sycl/set_rows.hpp +8 -0
  91. package/cpp/llama.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp +711 -292
  92. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp +58 -7
  93. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp +28 -23
  94. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.comp +14 -9
  95. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp +38 -32
  96. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp +32 -27
  97. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_split_k_reduce.comp +44 -12
  98. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/geglu.comp +13 -0
  99. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/geglu_erf.comp +27 -0
  100. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/geglu_quick.comp +11 -0
  101. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/gelu_erf.comp +39 -0
  102. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/glu_head.comp +15 -0
  103. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/glu_main.comp +29 -0
  104. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp +128 -72
  105. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp +38 -9
  106. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/reglu.comp +9 -0
  107. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm.comp +12 -3
  108. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/roll.comp +46 -0
  109. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/rope_multi.comp +7 -9
  110. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/rope_neox.comp +7 -9
  111. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/rope_norm.comp +7 -9
  112. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/scale.comp +1 -1
  113. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/soft_max.comp +20 -4
  114. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/swiglu.comp +9 -0
  115. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/upscale.comp +69 -5
  116. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp +23 -3
  117. package/cpp/llama.cpp/ggml/src/ggml.c +382 -61
  118. package/cpp/llama.cpp/ggml/src/gguf.cpp +8 -1
  119. package/cpp/llama.cpp/gguf-py/gguf/constants.py +209 -0
  120. package/cpp/llama.cpp/gguf-py/gguf/gguf_writer.py +8 -2
  121. package/cpp/llama.cpp/gguf-py/gguf/tensor_mapping.py +73 -21
  122. package/cpp/llama.cpp/gguf-py/gguf/vocab.py +12 -3
  123. package/cpp/llama.cpp/include/llama.h +0 -40
  124. package/cpp/llama.cpp/src/llama-arch.cpp +210 -3
  125. package/cpp/llama.cpp/src/llama-arch.h +18 -1
  126. package/cpp/llama.cpp/src/llama-batch.cpp +27 -1
  127. package/cpp/llama.cpp/src/llama-batch.h +8 -1
  128. package/cpp/llama.cpp/src/llama-chat.cpp +15 -0
  129. package/cpp/llama.cpp/src/llama-chat.h +1 -0
  130. package/cpp/llama.cpp/src/llama-graph.cpp +119 -184
  131. package/cpp/llama.cpp/src/llama-graph.h +47 -60
  132. package/cpp/llama.cpp/src/llama-hparams.cpp +7 -1
  133. package/cpp/llama.cpp/src/llama-hparams.h +3 -0
  134. package/cpp/llama.cpp/src/llama-kv-cache-unified-iswa.cpp +28 -18
  135. package/cpp/llama.cpp/src/llama-kv-cache-unified-iswa.h +4 -2
  136. package/cpp/llama.cpp/src/llama-kv-cache-unified.cpp +214 -65
  137. package/cpp/llama.cpp/src/llama-kv-cache-unified.h +62 -24
  138. package/cpp/llama.cpp/src/llama-kv-cells.h +62 -10
  139. package/cpp/llama.cpp/src/llama-memory-hybrid.cpp +9 -4
  140. package/cpp/llama.cpp/src/llama-memory-hybrid.h +3 -1
  141. package/cpp/llama.cpp/src/llama-memory-recurrent.cpp +20 -10
  142. package/cpp/llama.cpp/src/llama-memory.cpp +17 -0
  143. package/cpp/llama.cpp/src/llama-memory.h +3 -0
  144. package/cpp/llama.cpp/src/llama-model.cpp +2530 -685
  145. package/cpp/llama.cpp/src/llama-model.h +18 -0
  146. package/cpp/llama.cpp/src/llama-quant.cpp +1 -0
  147. package/cpp/llama.cpp/src/llama-vocab.cpp +13 -2
  148. package/cpp/llama.cpp/src/llama-vocab.h +41 -0
  149. package/ios/include/chat.h +2 -0
  150. package/ios/include/common.h +4 -0
  151. package/ios/include/llama.h +0 -40
  152. package/ios/libs/llama.xcframework/Info.plist +19 -19
  153. package/ios/libs/llama.xcframework/ios-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  154. package/ios/libs/llama.xcframework/ios-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5055 -4886
  155. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/ggml-backend.h +1 -1
  156. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/ggml.h +173 -10
  157. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/llama.h +0 -40
  158. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/llama +0 -0
  159. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  160. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5030 -4861
  161. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3889 -3764
  162. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/ggml-backend.h +1 -1
  163. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +173 -10
  164. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/llama.h +0 -40
  165. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/llama +0 -0
  166. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  167. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5030 -4861
  168. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3891 -3766
  169. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/ggml-backend.h +1 -1
  170. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/ggml.h +173 -10
  171. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/llama.h +0 -40
  172. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/ggml-backend.h +1 -1
  173. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/ggml.h +173 -10
  174. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/llama.h +0 -40
  175. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/llama +0 -0
  176. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/ggml-backend.h +1 -1
  177. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/ggml.h +173 -10
  178. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/llama.h +0 -40
  179. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/llama +0 -0
  180. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/llama +0 -0
  181. package/ios/libs/llama.xcframework/tvos-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  182. package/ios/libs/llama.xcframework/tvos-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5059 -4890
  183. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/ggml-backend.h +1 -1
  184. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/ggml.h +173 -10
  185. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/llama.h +0 -40
  186. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/llama +0 -0
  187. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  188. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5030 -4861
  189. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3889 -3764
  190. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/ggml-backend.h +1 -1
  191. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +173 -10
  192. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/llama.h +0 -40
  193. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/llama +0 -0
  194. package/ios/libs/llama.xcframework/xros-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  195. package/ios/libs/llama.xcframework/xros-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5091 -4922
  196. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/ggml-backend.h +1 -1
  197. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/ggml.h +173 -10
  198. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/llama.h +0 -40
  199. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/llama +0 -0
  200. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  201. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +5066 -4897
  202. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3919 -3794
  203. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/ggml-backend.h +1 -1
  204. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +173 -10
  205. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/llama.h +0 -40
  206. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/llama +0 -0
  207. package/package.json +1 -1
  208. package/cpp/llama.cpp/ggml/include/ggml-kompute.h +0 -50
  209. package/cpp/llama.cpp/ggml/src/ggml-kompute/CMakeLists.txt +0 -166
  210. package/cpp/llama.cpp/ggml/src/ggml-kompute/ggml-kompute.cpp +0 -2251
  211. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/common.comp +0 -112
  212. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_add.comp +0 -58
  213. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_addrow.comp +0 -25
  214. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f16.comp +0 -52
  215. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f32.comp +0 -52
  216. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f16.comp +0 -52
  217. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f32.comp +0 -52
  218. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_diagmask.comp +0 -30
  219. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_gelu.comp +0 -22
  220. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows.comp +0 -17
  221. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f16.comp +0 -31
  222. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f32.comp +0 -31
  223. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_0.comp +0 -38
  224. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_1.comp +0 -39
  225. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q6_k.comp +0 -44
  226. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul.comp +0 -52
  227. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_f16.comp +0 -69
  228. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_mat_f32.comp +0 -51
  229. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_0.comp +0 -33
  230. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_1.comp +0 -35
  231. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_k.comp +0 -140
  232. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q6_k.comp +0 -106
  233. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q8_0.comp +0 -73
  234. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n.comp +0 -52
  235. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n_pre.comp +0 -28
  236. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_norm.comp +0 -84
  237. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_relu.comp +0 -21
  238. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_rmsnorm.comp +0 -53
  239. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f16.comp +0 -52
  240. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f32.comp +0 -52
  241. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f16.comp +0 -52
  242. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f32.comp +0 -52
  243. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_scale.comp +0 -19
  244. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_scale_8.comp +0 -23
  245. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_silu.comp +0 -22
  246. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/op_softmax.comp +0 -72
  247. package/cpp/llama.cpp/ggml/src/ggml-kompute/kompute-shaders/rope_common.comp +0 -71
package/cpp/llama.cpp/src/llama-model.h CHANGED
@@ -32,16 +32,21 @@ enum llm_type {
32
32
  LLM_TYPE_190M,
33
33
  LLM_TYPE_220M,
34
34
  LLM_TYPE_250M,
35
+ LLM_TYPE_256M,
35
36
  LLM_TYPE_270M,
36
37
  LLM_TYPE_335M,
38
+ LLM_TYPE_350M,
37
39
  LLM_TYPE_410M,
38
40
  LLM_TYPE_450M,
39
41
  LLM_TYPE_475M,
42
+ LLM_TYPE_700M,
40
43
  LLM_TYPE_770M,
41
44
  LLM_TYPE_780M,
45
+ LLM_TYPE_0_3B,
42
46
  LLM_TYPE_0_5B,
43
47
  LLM_TYPE_0_6B,
44
48
  LLM_TYPE_1B,
49
+ LLM_TYPE_1_2B,
45
50
  LLM_TYPE_1_3B,
46
51
  LLM_TYPE_1_4B,
47
52
  LLM_TYPE_1_5B,
@@ -93,6 +98,7 @@ enum llm_type {
93
98
  LLM_TYPE_57B_A14B,
94
99
  LLM_TYPE_17B_16E, // llama4 Scout
95
100
  LLM_TYPE_17B_128E, // llama4 Maverick
101
+ LLM_TYPE_A13B,
96
102
  LLM_TYPE_30B_A3B,
97
103
  LLM_TYPE_235B_A22B,
98
104
  LLM_TYPE_E2B,
@@ -152,6 +158,12 @@ struct llama_layer_convnext {
152
158
  struct ggml_tensor * gamma = nullptr;
153
159
  };
154
160
 
161
+ struct llama_layer_shortconv {
162
+ struct ggml_tensor * in_proj = nullptr;
163
+ struct ggml_tensor * conv = nullptr;
164
+ struct ggml_tensor * out_proj = nullptr;
165
+ };
166
+
155
167
  struct llama_layer {
156
168
  // normalization
157
169
  struct ggml_tensor * attn_norm = nullptr;
@@ -171,6 +183,10 @@ struct llama_layer {
171
183
  struct ggml_tensor * ffn_sub_norm = nullptr;
172
184
  struct ggml_tensor * attn_norm_cross = nullptr;
173
185
  struct ggml_tensor * attn_norm_enc = nullptr;
186
+ struct ggml_tensor * ssm_norm = nullptr;
187
+ struct ggml_tensor * ssm_dt_norm = nullptr;
188
+ struct ggml_tensor * ssm_b_norm = nullptr;
189
+ struct ggml_tensor * ssm_c_norm = nullptr;
174
190
 
175
191
  // attention
176
192
  struct ggml_tensor * wq = nullptr;
@@ -334,6 +350,8 @@ struct llama_layer {
334
350
  struct llama_layer_posnet posnet;
335
351
 
336
352
  struct llama_layer_convnext convnext;
353
+
354
+ struct llama_layer_shortconv shortconv;
337
355
  };
338
356
 
339
357
  struct llama_model {
package/cpp/llama.cpp/src/llama-quant.cpp CHANGED
@@ -844,6 +844,7 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
844
844
  // do not quantize Mamba's small yet 2D weights
845
845
  // NOTE: can't use LLM_TN here because the layer number is not known
846
846
  quantize &= name.find("ssm_conv1d.weight") == std::string::npos;
847
+ quantize &= name.find("shortconv.conv.weight") == std::string::npos;
847
848
 
848
849
  // do not quantize RWKV's small yet 2D weights
849
850
  quantize &= name.find("time_mix_first.weight") == std::string::npos;
package/cpp/llama.cpp/src/llama-vocab.cpp CHANGED
@@ -351,6 +351,7 @@ struct llm_tokenizer_bpe : llm_tokenizer {
351
351
  break;
352
352
  case LLAMA_VOCAB_PRE_TYPE_STABLELM2:
353
353
  case LLAMA_VOCAB_PRE_TYPE_QWEN2:
354
+ case LLAMA_VOCAB_PRE_TYPE_HUNYUAN:
354
355
  regex_exprs = {
355
356
  // original regex from tokenizer.json
356
357
  // "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?\\p{L}+|\\p{N}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+"
@@ -1522,7 +1523,10 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
1522
1523
  tokenizer_pre == "llama-v3" ||
1523
1524
  tokenizer_pre == "llama-bpe"||
1524
1525
  tokenizer_pre == "falcon3" ||
1525
- tokenizer_pre == "pixtral") {
1526
+ tokenizer_pre == "falcon-h1" ||
1527
+ tokenizer_pre == "pixtral" ||
1528
+ tokenizer_pre == "midm-2.0" ||
1529
+ tokenizer_pre == "lfm2") {
1526
1530
  pre_type = LLAMA_VOCAB_PRE_TYPE_LLAMA3;
1527
1531
  ignore_merges = true;
1528
1532
  add_bos = true;
@@ -1554,7 +1558,8 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
1554
1558
  tokenizer_pre == "jina-de" ||
1555
1559
  tokenizer_pre == "gigachat" ||
1556
1560
  tokenizer_pre == "jina-v2-es" ||
1557
- tokenizer_pre == "jina-v2-de") {
1561
+ tokenizer_pre == "jina-v2-de" ||
1562
+ tokenizer_pre == "a.x-4.0") {
1558
1563
  pre_type = LLAMA_VOCAB_PRE_TYPE_GPT2;
1559
1564
  } else if (
1560
1565
  tokenizer_pre == "jina-v1-en" ||
@@ -1656,6 +1661,10 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
1656
1661
  tokenizer_pre == "seed-coder") {
1657
1662
  pre_type = LLAMA_VOCAB_PRE_TYPE_SEED_CODER;
1658
1663
  clean_spaces = false;
1664
+ } else if (
1665
+ tokenizer_pre == "hunyuan") {
1666
+ pre_type = LLAMA_VOCAB_PRE_TYPE_HUNYUAN;
1667
+ clean_spaces = false;
1659
1668
  } else {
1660
1669
  throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
1661
1670
  }
@@ -1839,6 +1848,7 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
1839
1848
  || t.first == "<EOT>"
1840
1849
  || t.first == "_<EOT>"
1841
1850
  || t.first == "<｜end▁of▁sentence｜>" // DeepSeek
1851
+ || t.first == "<end_of_utterance>" // smoldocling
1842
1852
  ) {
1843
1853
  special_eot_id = t.second;
1844
1854
  if ((id_to_token[t.second].attr & LLAMA_TOKEN_ATTR_CONTROL) == 0) {
@@ -1998,6 +2008,7 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
1998
2008
  || t.first == "<EOT>"
1999
2009
  || t.first == "_<EOT>"
2000
2010
  || t.first == "<|end_of_text|>"
2011
+ || t.first == "<end_of_utterance>" // smoldocling
2001
2012
  ) {
2002
2013
  special_eog_ids.insert(t.second);
2003
2014
  if ((id_to_token[t.second].attr & LLAMA_TOKEN_ATTR_CONTROL) == 0) {
package/cpp/llama.cpp/src/llama-vocab.h CHANGED
@@ -6,6 +6,47 @@
6
6
  #include <vector>
7
7
  #include <memory>
8
8
 
9
+ // pre-tokenization types
10
+ enum llama_vocab_pre_type {
11
+ LLAMA_VOCAB_PRE_TYPE_DEFAULT = 0,
12
+ LLAMA_VOCAB_PRE_TYPE_LLAMA3 = 1,
13
+ LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_LLM = 2,
14
+ LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER = 3,
15
+ LLAMA_VOCAB_PRE_TYPE_FALCON = 4,
16
+ LLAMA_VOCAB_PRE_TYPE_MPT = 5,
17
+ LLAMA_VOCAB_PRE_TYPE_STARCODER = 6,
18
+ LLAMA_VOCAB_PRE_TYPE_GPT2 = 7,
19
+ LLAMA_VOCAB_PRE_TYPE_REFACT = 8,
20
+ LLAMA_VOCAB_PRE_TYPE_COMMAND_R = 9,
21
+ LLAMA_VOCAB_PRE_TYPE_STABLELM2 = 10,
22
+ LLAMA_VOCAB_PRE_TYPE_QWEN2 = 11,
23
+ LLAMA_VOCAB_PRE_TYPE_OLMO = 12,
24
+ LLAMA_VOCAB_PRE_TYPE_DBRX = 13,
25
+ LLAMA_VOCAB_PRE_TYPE_SMAUG = 14,
26
+ LLAMA_VOCAB_PRE_TYPE_PORO = 15,
27
+ LLAMA_VOCAB_PRE_TYPE_CHATGLM3 = 16,
28
+ LLAMA_VOCAB_PRE_TYPE_CHATGLM4 = 17,
29
+ LLAMA_VOCAB_PRE_TYPE_VIKING = 18,
30
+ LLAMA_VOCAB_PRE_TYPE_JAIS = 19,
31
+ LLAMA_VOCAB_PRE_TYPE_TEKKEN = 20,
32
+ LLAMA_VOCAB_PRE_TYPE_SMOLLM = 21,
33
+ LLAMA_VOCAB_PRE_TYPE_CODESHELL = 22,
34
+ LLAMA_VOCAB_PRE_TYPE_BLOOM = 23,
35
+ LLAMA_VOCAB_PRE_TYPE_GPT3_FINNISH = 24,
36
+ LLAMA_VOCAB_PRE_TYPE_EXAONE = 25,
37
+ LLAMA_VOCAB_PRE_TYPE_CHAMELEON = 26,
38
+ LLAMA_VOCAB_PRE_TYPE_MINERVA = 27,
39
+ LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM = 28,
40
+ LLAMA_VOCAB_PRE_TYPE_GPT4O = 29,
41
+ LLAMA_VOCAB_PRE_TYPE_SUPERBPE = 30,
42
+ LLAMA_VOCAB_PRE_TYPE_TRILLION = 31,
43
+ LLAMA_VOCAB_PRE_TYPE_BAILINGMOE = 32,
44
+ LLAMA_VOCAB_PRE_TYPE_LLAMA4 = 33,
45
+ LLAMA_VOCAB_PRE_TYPE_PIXTRAL = 34,
46
+ LLAMA_VOCAB_PRE_TYPE_SEED_CODER = 35,
47
+ LLAMA_VOCAB_PRE_TYPE_HUNYUAN = 36,
48
+ };
49
+
9
50
  struct LLM_KV;
10
51
  struct llama_model_loader;
11
52
 
package/ios/include/chat.h CHANGED
@@ -7,6 +7,7 @@
7
7
  #include <chrono>
8
8
  #include <string>
9
9
  #include <vector>
10
+ #include <map>
10
11
 
11
12
  struct common_chat_templates;
12
13
 
@@ -125,6 +126,7 @@ struct common_chat_templates_inputs {
125
126
  common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE;
126
127
  bool enable_thinking = true;
127
128
  std::chrono::system_clock::time_point now = std::chrono::system_clock::now();
129
+ std::map<std::string, std::string> chat_template_kwargs;
128
130
  };
129
131
 
130
132
  struct common_chat_params {
package/ios/include/common.h CHANGED
@@ -8,6 +8,7 @@
8
8
  #include <string>
9
9
  #include <string_view>
10
10
  #include <vector>
11
+ #include <map>
11
12
  #include <sstream>
12
13
 
13
14
  #ifdef _WIN32
@@ -369,6 +370,7 @@ struct common_params {
369
370
 
370
371
  std::string hostname = "127.0.0.1";
371
372
  std::string public_path = ""; // NOLINT
373
+ std::string api_prefix = ""; // NOLINT
372
374
  std::string chat_template = ""; // NOLINT
373
375
  bool use_jinja = false; // NOLINT
374
376
  bool enable_chat_template = true;
@@ -381,6 +383,8 @@ struct common_params {
381
383
  std::string ssl_file_key = ""; // NOLINT
382
384
  std::string ssl_file_cert = ""; // NOLINT
383
385
 
386
+ std::map<std::string, std::string> default_template_kwargs;
387
+
384
388
  // "advanced" endpoints are disabled by default for better security
385
389
  bool webui = true;
386
390
  bool endpoint_slots = false;
package/ios/include/llama.h CHANGED
@@ -79,46 +79,6 @@ extern "C" {
79
79
  LLAMA_VOCAB_TYPE_RWKV = 5, // RWKV tokenizer based on greedy tokenization
80
80
  };
81
81
 
82
- // pre-tokenization types
83
- enum llama_vocab_pre_type {
84
- LLAMA_VOCAB_PRE_TYPE_DEFAULT = 0,
85
- LLAMA_VOCAB_PRE_TYPE_LLAMA3 = 1,
86
- LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_LLM = 2,
87
- LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER = 3,
88
- LLAMA_VOCAB_PRE_TYPE_FALCON = 4,
89
- LLAMA_VOCAB_PRE_TYPE_MPT = 5,
90
- LLAMA_VOCAB_PRE_TYPE_STARCODER = 6,
91
- LLAMA_VOCAB_PRE_TYPE_GPT2 = 7,
92
- LLAMA_VOCAB_PRE_TYPE_REFACT = 8,
93
- LLAMA_VOCAB_PRE_TYPE_COMMAND_R = 9,
94
- LLAMA_VOCAB_PRE_TYPE_STABLELM2 = 10,
95
- LLAMA_VOCAB_PRE_TYPE_QWEN2 = 11,
96
- LLAMA_VOCAB_PRE_TYPE_OLMO = 12,
97
- LLAMA_VOCAB_PRE_TYPE_DBRX = 13,
98
- LLAMA_VOCAB_PRE_TYPE_SMAUG = 14,
99
- LLAMA_VOCAB_PRE_TYPE_PORO = 15,
100
- LLAMA_VOCAB_PRE_TYPE_CHATGLM3 = 16,
101
- LLAMA_VOCAB_PRE_TYPE_CHATGLM4 = 17,
102
- LLAMA_VOCAB_PRE_TYPE_VIKING = 18,
103
- LLAMA_VOCAB_PRE_TYPE_JAIS = 19,
104
- LLAMA_VOCAB_PRE_TYPE_TEKKEN = 20,
105
- LLAMA_VOCAB_PRE_TYPE_SMOLLM = 21,
106
- LLAMA_VOCAB_PRE_TYPE_CODESHELL = 22,
107
- LLAMA_VOCAB_PRE_TYPE_BLOOM = 23,
108
- LLAMA_VOCAB_PRE_TYPE_GPT3_FINNISH = 24,
109
- LLAMA_VOCAB_PRE_TYPE_EXAONE = 25,
110
- LLAMA_VOCAB_PRE_TYPE_CHAMELEON = 26,
111
- LLAMA_VOCAB_PRE_TYPE_MINERVA = 27,
112
- LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM = 28,
113
- LLAMA_VOCAB_PRE_TYPE_GPT4O = 29,
114
- LLAMA_VOCAB_PRE_TYPE_SUPERBPE = 30,
115
- LLAMA_VOCAB_PRE_TYPE_TRILLION = 31,
116
- LLAMA_VOCAB_PRE_TYPE_BAILINGMOE = 32,
117
- LLAMA_VOCAB_PRE_TYPE_LLAMA4 = 33,
118
- LLAMA_VOCAB_PRE_TYPE_PIXTRAL = 34,
119
- LLAMA_VOCAB_PRE_TYPE_SEED_CODER = 35,
120
- };
121
-
122
82
  enum llama_rope_type {
123
83
  LLAMA_ROPE_TYPE_NONE = -1,
124
84
  LLAMA_ROPE_TYPE_NORM = 0,
package/ios/libs/llama.xcframework/Info.plist CHANGED
@@ -10,7 +10,7 @@
10
10
  <key>DebugSymbolsPath</key>
11
11
  <string>dSYMs</string>
12
12
  <key>LibraryIdentifier</key>
13
- <string>ios-arm64_x86_64-simulator</string>
13
+ <string>xros-arm64_x86_64-simulator</string>
14
14
  <key>LibraryPath</key>
15
15
  <string>llama.framework</string>
16
16
  <key>SupportedArchitectures</key>
@@ -19,7 +19,7 @@
19
19
  <string>x86_64</string>
20
20
  </array>
21
21
  <key>SupportedPlatform</key>
22
- <string>ios</string>
22
+ <string>xros</string>
23
23
  <key>SupportedPlatformVariant</key>
24
24
  <string>simulator</string>
25
25
  </dict>
@@ -29,18 +29,15 @@
29
29
  <key>DebugSymbolsPath</key>
30
30
  <string>dSYMs</string>
31
31
  <key>LibraryIdentifier</key>
32
- <string>xros-arm64_x86_64-simulator</string>
32
+ <string>ios-arm64</string>
33
33
  <key>LibraryPath</key>
34
34
  <string>llama.framework</string>
35
35
  <key>SupportedArchitectures</key>
36
36
  <array>
37
37
  <string>arm64</string>
38
- <string>x86_64</string>
39
38
  </array>
40
39
  <key>SupportedPlatform</key>
41
- <string>xros</string>
42
- <key>SupportedPlatformVariant</key>
43
- <string>simulator</string>
40
+ <string>ios</string>
44
41
  </dict>
45
42
  <dict>
46
43
  <key>BinaryPath</key>
@@ -48,7 +45,7 @@
48
45
  <key>DebugSymbolsPath</key>
49
46
  <string>dSYMs</string>
50
47
  <key>LibraryIdentifier</key>
51
- <string>ios-arm64</string>
48
+ <string>tvos-arm64</string>
52
49
  <key>LibraryPath</key>
53
50
  <string>llama.framework</string>
54
51
  <key>SupportedArchitectures</key>
@@ -56,7 +53,7 @@
56
53
  <string>arm64</string>
57
54
  </array>
58
55
  <key>SupportedPlatform</key>
59
- <string>ios</string>
56
+ <string>tvos</string>
60
57
  </dict>
61
58
  <dict>
62
59
  <key>BinaryPath</key>
@@ -64,7 +61,7 @@
64
61
  <key>DebugSymbolsPath</key>
65
62
  <string>dSYMs</string>
66
63
  <key>LibraryIdentifier</key>
67
- <string>tvos-arm64_x86_64-simulator</string>
64
+ <string>ios-arm64_x86_64-simulator</string>
68
65
  <key>LibraryPath</key>
69
66
  <string>llama.framework</string>
70
67
  <key>SupportedArchitectures</key>
@@ -73,42 +70,42 @@
73
70
  <string>x86_64</string>
74
71
  </array>
75
72
  <key>SupportedPlatform</key>
76
- <string>tvos</string>
73
+ <string>ios</string>
77
74
  <key>SupportedPlatformVariant</key>
78
75
  <string>simulator</string>
79
76
  </dict>
80
77
  <dict>
81
78
  <key>BinaryPath</key>
82
- <string>llama.framework/llama</string>
79
+ <string>llama.framework/Versions/A/llama</string>
83
80
  <key>DebugSymbolsPath</key>
84
81
  <string>dSYMs</string>
85
82
  <key>LibraryIdentifier</key>
86
- <string>xros-arm64</string>
83
+ <string>macos-arm64_x86_64</string>
87
84
  <key>LibraryPath</key>
88
85
  <string>llama.framework</string>
89
86
  <key>SupportedArchitectures</key>
90
87
  <array>
91
88
  <string>arm64</string>
89
+ <string>x86_64</string>
92
90
  </array>
93
91
  <key>SupportedPlatform</key>
94
- <string>xros</string>
92
+ <string>macos</string>
95
93
  </dict>
96
94
  <dict>
97
95
  <key>BinaryPath</key>
98
- <string>llama.framework/Versions/A/llama</string>
96
+ <string>llama.framework/llama</string>
99
97
  <key>DebugSymbolsPath</key>
100
98
  <string>dSYMs</string>
101
99
  <key>LibraryIdentifier</key>
102
- <string>macos-arm64_x86_64</string>
100
+ <string>xros-arm64</string>
103
101
  <key>LibraryPath</key>
104
102
  <string>llama.framework</string>
105
103
  <key>SupportedArchitectures</key>
106
104
  <array>
107
105
  <string>arm64</string>
108
- <string>x86_64</string>
109
106
  </array>
110
107
  <key>SupportedPlatform</key>
111
- <string>macos</string>
108
+ <string>xros</string>
112
109
  </dict>
113
110
  <dict>
114
111
  <key>BinaryPath</key>
@@ -116,15 +113,18 @@
116
113
  <key>DebugSymbolsPath</key>
117
114
  <string>dSYMs</string>
118
115
  <key>LibraryIdentifier</key>
119
- <string>tvos-arm64</string>
116
+ <string>tvos-arm64_x86_64-simulator</string>
120
117
  <key>LibraryPath</key>
121
118
  <string>llama.framework</string>
122
119
  <key>SupportedArchitectures</key>
123
120
  <array>
124
121
  <string>arm64</string>
122
+ <string>x86_64</string>
125
123
  </array>
126
124
  <key>SupportedPlatform</key>
127
125
  <string>tvos</string>
126
+ <key>SupportedPlatformVariant</key>
127
+ <string>simulator</string>
128
128
  </dict>
129
129
  </array>
130
130
  <key>CFBundlePackageType</key>