@novastera-oss/llamarn 0.2.1 → 0.2.3

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 (268) hide show
  1. package/README.md +80 -14
  2. package/RNLlamaCpp.podspec +10 -3
  3. package/android/CMakeLists.txt +8 -0
  4. package/android/src/main/cpp/include/llama.h +62 -125
  5. package/android/src/main/jniLibs/arm64-v8a/libggml-base.so +0 -0
  6. package/android/src/main/jniLibs/arm64-v8a/libggml-cpu.so +0 -0
  7. package/android/src/main/jniLibs/arm64-v8a/libggml.so +0 -0
  8. package/android/src/main/jniLibs/arm64-v8a/libllama.so +0 -0
  9. package/android/src/main/jniLibs/x86_64/libggml-base.so +0 -0
  10. package/android/src/main/jniLibs/x86_64/libggml-cpu.so +0 -0
  11. package/android/src/main/jniLibs/x86_64/libggml.so +0 -0
  12. package/android/src/main/jniLibs/x86_64/libllama.so +0 -0
  13. package/cpp/PureCppImpl.cpp +9 -27
  14. package/cpp/SystemUtils.h +2 -2
  15. package/cpp/build-info.cpp +2 -2
  16. package/cpp/llama.cpp/README.md +11 -3
  17. package/cpp/llama.cpp/build-xcframework.sh +1 -0
  18. package/cpp/llama.cpp/common/CMakeLists.txt +8 -2
  19. package/cpp/llama.cpp/common/arg.cpp +153 -113
  20. package/cpp/llama.cpp/common/chat-parser.cpp +379 -0
  21. package/cpp/llama.cpp/common/chat-parser.h +117 -0
  22. package/cpp/llama.cpp/common/chat.cpp +847 -699
  23. package/cpp/llama.cpp/common/chat.h +73 -6
  24. package/cpp/llama.cpp/common/common.cpp +50 -82
  25. package/cpp/llama.cpp/common/common.h +21 -17
  26. package/cpp/llama.cpp/common/json-partial.cpp +255 -0
  27. package/cpp/llama.cpp/common/json-partial.h +37 -0
  28. package/cpp/llama.cpp/common/minja/chat-template.hpp +9 -5
  29. package/cpp/llama.cpp/common/minja/minja.hpp +69 -36
  30. package/cpp/llama.cpp/common/regex-partial.cpp +204 -0
  31. package/cpp/llama.cpp/common/regex-partial.h +56 -0
  32. package/cpp/llama.cpp/common/sampling.cpp +7 -8
  33. package/cpp/llama.cpp/convert_hf_to_gguf.py +453 -118
  34. package/cpp/llama.cpp/convert_hf_to_gguf_update.py +120 -68
  35. package/cpp/llama.cpp/ggml/CMakeLists.txt +2 -1
  36. package/cpp/llama.cpp/ggml/cmake/common.cmake +25 -0
  37. package/cpp/llama.cpp/ggml/include/ggml-opt.h +49 -28
  38. package/cpp/llama.cpp/ggml/include/ggml.h +26 -7
  39. package/cpp/llama.cpp/ggml/src/CMakeLists.txt +16 -10
  40. package/cpp/llama.cpp/ggml/src/ggml-backend.cpp +4 -1
  41. package/cpp/llama.cpp/ggml/src/ggml-cann/CMakeLists.txt +1 -0
  42. package/cpp/llama.cpp/ggml/src/ggml-cann/acl_tensor.cpp +2 -0
  43. package/cpp/llama.cpp/ggml/src/ggml-cann/aclnn_ops.cpp +604 -0
  44. package/cpp/llama.cpp/ggml/src/ggml-cann/aclnn_ops.h +42 -0
  45. package/cpp/llama.cpp/ggml/src/ggml-cann/ggml-cann.cpp +54 -2
  46. package/cpp/llama.cpp/ggml/src/ggml-cpu/CMakeLists.txt +50 -51
  47. package/cpp/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp +2 -2
  48. package/cpp/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-impl.h +5 -9
  49. package/cpp/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-quants.c +779 -19
  50. package/cpp/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c +22 -0
  51. package/cpp/llama.cpp/ggml/src/ggml-cpu/kleidiai/kernels.cpp +88 -5
  52. package/cpp/llama.cpp/ggml/src/ggml-cpu/kleidiai/kernels.h +47 -12
  53. package/cpp/llama.cpp/ggml/src/ggml-cpu/kleidiai/kleidiai.cpp +264 -69
  54. package/cpp/llama.cpp/ggml/src/ggml-cpu/ops.cpp +322 -100
  55. package/cpp/llama.cpp/ggml/src/ggml-cpu/simd-mappings.h +117 -1
  56. package/cpp/llama.cpp/ggml/src/ggml-cpu/vec.cpp +85 -16
  57. package/cpp/llama.cpp/ggml/src/ggml-cpu/vec.h +220 -49
  58. package/cpp/llama.cpp/ggml/src/ggml-cuda/acc.cu +40 -26
  59. package/cpp/llama.cpp/ggml/src/ggml-cuda/common.cuh +1 -1
  60. package/cpp/llama.cpp/ggml/src/ggml-cuda/cpy.cu +11 -1
  61. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-common.cuh +15 -7
  62. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-mma-f16.cuh +266 -64
  63. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-vec-f16.cuh +49 -4
  64. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn-vec-f32.cuh +48 -4
  65. package/cpp/llama.cpp/ggml/src/ggml-cuda/fattn.cu +2 -1
  66. package/cpp/llama.cpp/ggml/src/ggml-cuda/ggml-cuda.cu +5 -1
  67. package/cpp/llama.cpp/ggml/src/ggml-cuda/mmq.cu +2 -0
  68. package/cpp/llama.cpp/ggml/src/ggml-cuda/quantize.cu +7 -6
  69. package/cpp/llama.cpp/ggml/src/ggml-cuda/sum.cu +1 -1
  70. package/cpp/llama.cpp/ggml/src/ggml-cuda/unary.cu +10 -0
  71. package/cpp/llama.cpp/ggml/src/ggml-cuda/unary.cuh +2 -0
  72. package/cpp/llama.cpp/ggml/src/ggml-impl.h +1 -1
  73. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal-impl.h +4 -0
  74. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal.m +99 -17
  75. package/cpp/llama.cpp/ggml/src/ggml-metal/ggml-metal.metal +200 -2
  76. package/cpp/llama.cpp/ggml/src/ggml-musa/CMakeLists.txt +8 -2
  77. package/cpp/llama.cpp/ggml/src/ggml-musa/mudnn.cu +112 -0
  78. package/cpp/llama.cpp/ggml/src/ggml-musa/mudnn.cuh +12 -0
  79. package/cpp/llama.cpp/ggml/src/ggml-opencl/CMakeLists.txt +6 -0
  80. package/cpp/llama.cpp/ggml/src/ggml-opencl/ggml-opencl.cpp +972 -178
  81. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/argsort.cl +86 -0
  82. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/div.cl +72 -0
  83. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/group_norm.cl +72 -0
  84. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/sigmoid.cl +29 -0
  85. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/sub.cl +72 -0
  86. package/cpp/llama.cpp/ggml/src/ggml-opencl/kernels/sum_rows.cl +39 -0
  87. package/cpp/llama.cpp/ggml/src/ggml-opt.cpp +373 -190
  88. package/cpp/llama.cpp/ggml/src/ggml-sycl/CMakeLists.txt +29 -23
  89. package/cpp/llama.cpp/ggml/src/ggml-sycl/binbcast.cpp +5 -10
  90. package/cpp/llama.cpp/ggml/src/ggml-sycl/common.hpp +101 -5
  91. package/cpp/llama.cpp/ggml/src/ggml-sycl/concat.cpp +31 -33
  92. package/cpp/llama.cpp/ggml/src/ggml-sycl/conv.cpp +1 -0
  93. package/cpp/llama.cpp/ggml/src/ggml-sycl/convert.cpp +29 -2
  94. package/cpp/llama.cpp/ggml/src/ggml-sycl/cpy.cpp +4 -5
  95. package/cpp/llama.cpp/ggml/src/ggml-sycl/dequantize.hpp +59 -21
  96. package/cpp/llama.cpp/ggml/src/ggml-sycl/dmmv.cpp +9 -1
  97. package/cpp/llama.cpp/ggml/src/ggml-sycl/element_wise.cpp +84 -72
  98. package/cpp/llama.cpp/ggml/src/ggml-sycl/element_wise.hpp +2 -0
  99. package/cpp/llama.cpp/ggml/src/ggml-sycl/gemm.hpp +37 -8
  100. package/cpp/llama.cpp/ggml/src/ggml-sycl/getrows.cpp +1 -3
  101. package/cpp/llama.cpp/ggml/src/ggml-sycl/ggml-sycl.cpp +324 -129
  102. package/cpp/llama.cpp/ggml/src/ggml-sycl/gla.cpp +1 -0
  103. package/cpp/llama.cpp/ggml/src/ggml-sycl/mmvq.cpp +31 -2
  104. package/cpp/llama.cpp/ggml/src/ggml-sycl/norm.cpp +95 -68
  105. package/cpp/llama.cpp/ggml/src/ggml-sycl/outprod.cpp +1 -0
  106. package/cpp/llama.cpp/ggml/src/ggml-sycl/quants.hpp +22 -0
  107. package/cpp/llama.cpp/ggml/src/ggml-sycl/rope.cpp +1 -2
  108. package/cpp/llama.cpp/ggml/src/ggml-sycl/softmax.cpp +1 -4
  109. package/cpp/llama.cpp/ggml/src/ggml-sycl/tsembd.cpp +2 -3
  110. package/cpp/llama.cpp/ggml/src/ggml-sycl/vecdotq.hpp +69 -43
  111. package/cpp/llama.cpp/ggml/src/ggml-sycl/wkv.cpp +2 -14
  112. package/cpp/llama.cpp/ggml/src/ggml-vulkan/CMakeLists.txt +81 -91
  113. package/cpp/llama.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp +432 -181
  114. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/CMakeLists.txt +17 -0
  115. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq1_m.comp +1 -1
  116. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp +6 -152
  117. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.comp +162 -0
  118. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp +360 -0
  119. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp +2 -118
  120. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp +1 -1
  121. package/cpp/llama.cpp/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp +12 -1
  122. package/cpp/llama.cpp/ggml/src/ggml.c +107 -36
  123. package/cpp/llama.cpp/ggml/src/gguf.cpp +33 -33
  124. package/cpp/llama.cpp/gguf-py/gguf/constants.py +100 -15
  125. package/cpp/llama.cpp/gguf-py/gguf/gguf_reader.py +1 -1
  126. package/cpp/llama.cpp/gguf-py/gguf/gguf_writer.py +44 -12
  127. package/cpp/llama.cpp/gguf-py/gguf/scripts/gguf_editor_gui.py +21 -10
  128. package/cpp/llama.cpp/gguf-py/gguf/scripts/gguf_new_metadata.py +5 -2
  129. package/cpp/llama.cpp/gguf-py/gguf/tensor_mapping.py +128 -31
  130. package/cpp/llama.cpp/gguf-py/gguf/utility.py +1 -1
  131. package/cpp/llama.cpp/gguf-py/pyproject.toml +1 -1
  132. package/cpp/llama.cpp/include/llama.h +62 -125
  133. package/cpp/llama.cpp/models/ggml-vocab-bert-bge.gguf.inp +1 -1
  134. package/cpp/llama.cpp/models/ggml-vocab-bert-bge.gguf.out +1 -1
  135. package/cpp/llama.cpp/models/ggml-vocab-command-r.gguf.inp +1 -1
  136. package/cpp/llama.cpp/models/ggml-vocab-command-r.gguf.out +1 -1
  137. package/cpp/llama.cpp/models/ggml-vocab-deepseek-coder.gguf.inp +1 -1
  138. package/cpp/llama.cpp/models/ggml-vocab-deepseek-coder.gguf.out +1 -1
  139. package/cpp/llama.cpp/models/ggml-vocab-deepseek-llm.gguf.inp +1 -1
  140. package/cpp/llama.cpp/models/ggml-vocab-deepseek-llm.gguf.out +1 -1
  141. package/cpp/llama.cpp/models/ggml-vocab-falcon.gguf.inp +1 -1
  142. package/cpp/llama.cpp/models/ggml-vocab-falcon.gguf.out +1 -1
  143. package/cpp/llama.cpp/models/ggml-vocab-gpt-2.gguf.inp +1 -1
  144. package/cpp/llama.cpp/models/ggml-vocab-gpt-2.gguf.out +1 -1
  145. package/cpp/llama.cpp/models/ggml-vocab-llama-bpe.gguf.inp +1 -1
  146. package/cpp/llama.cpp/models/ggml-vocab-llama-bpe.gguf.out +1 -1
  147. package/cpp/llama.cpp/models/ggml-vocab-llama-spm.gguf.inp +1 -1
  148. package/cpp/llama.cpp/models/ggml-vocab-llama-spm.gguf.out +1 -1
  149. package/cpp/llama.cpp/models/ggml-vocab-mpt.gguf.inp +1 -1
  150. package/cpp/llama.cpp/models/ggml-vocab-mpt.gguf.out +1 -1
  151. package/cpp/llama.cpp/models/ggml-vocab-nomic-bert-moe.gguf +0 -0
  152. package/cpp/llama.cpp/models/ggml-vocab-phi-3.gguf.inp +1 -1
  153. package/cpp/llama.cpp/models/ggml-vocab-phi-3.gguf.out +1 -1
  154. package/cpp/llama.cpp/models/ggml-vocab-qwen2.gguf.inp +1 -1
  155. package/cpp/llama.cpp/models/ggml-vocab-qwen2.gguf.out +1 -1
  156. package/cpp/llama.cpp/models/ggml-vocab-refact.gguf.inp +1 -1
  157. package/cpp/llama.cpp/models/ggml-vocab-refact.gguf.out +1 -1
  158. package/cpp/llama.cpp/models/ggml-vocab-starcoder.gguf.inp +1 -1
  159. package/cpp/llama.cpp/models/ggml-vocab-starcoder.gguf.out +1 -1
  160. package/cpp/llama.cpp/models/templates/Qwen-QwQ-32B.jinja +62 -0
  161. package/cpp/llama.cpp/models/templates/Qwen-Qwen3-0.6B.jinja +85 -0
  162. package/cpp/llama.cpp/models/templates/README.md +2 -0
  163. package/cpp/llama.cpp/requirements/requirements-convert_hf_to_gguf.txt +5 -1
  164. package/cpp/llama.cpp/requirements/requirements-convert_hf_to_gguf_update.txt +5 -1
  165. package/cpp/llama.cpp/requirements/requirements-convert_lora_to_gguf.txt +2 -0
  166. package/cpp/llama.cpp/requirements/requirements-gguf_editor_gui.txt +1 -1
  167. package/cpp/llama.cpp/src/CMakeLists.txt +2 -0
  168. package/cpp/llama.cpp/src/llama-arch.cpp +6 -0
  169. package/cpp/llama.cpp/src/llama-arch.h +2 -0
  170. package/cpp/llama.cpp/src/llama-batch.cpp +3 -1
  171. package/cpp/llama.cpp/src/llama-context.cpp +340 -123
  172. package/cpp/llama.cpp/src/llama-context.h +30 -0
  173. package/cpp/llama.cpp/src/llama-cparams.cpp +4 -0
  174. package/cpp/llama.cpp/src/llama-cparams.h +2 -0
  175. package/cpp/llama.cpp/src/llama-grammar.cpp +12 -2
  176. package/cpp/llama.cpp/src/llama-graph.cpp +157 -247
  177. package/cpp/llama.cpp/src/llama-graph.h +52 -7
  178. package/cpp/llama.cpp/src/llama-hparams.cpp +17 -1
  179. package/cpp/llama.cpp/src/llama-hparams.h +37 -5
  180. package/cpp/llama.cpp/src/llama-kv-cache.cpp +742 -481
  181. package/cpp/llama.cpp/src/llama-kv-cache.h +196 -99
  182. package/cpp/llama.cpp/src/llama-kv-cells.h +379 -0
  183. package/cpp/llama.cpp/src/llama-memory.h +4 -3
  184. package/cpp/llama.cpp/src/llama-model-loader.cpp +22 -17
  185. package/cpp/llama.cpp/src/llama-model-saver.cpp +281 -0
  186. package/cpp/llama.cpp/src/llama-model-saver.h +37 -0
  187. package/cpp/llama.cpp/src/llama-model.cpp +529 -172
  188. package/cpp/llama.cpp/src/llama-model.h +6 -1
  189. package/cpp/llama.cpp/src/llama-quant.cpp +15 -13
  190. package/cpp/llama.cpp/src/llama-sampling.cpp +2 -2
  191. package/cpp/llama.cpp/src/llama-vocab.cpp +35 -8
  192. package/cpp/llama.cpp/src/llama-vocab.h +6 -0
  193. package/cpp/llama.cpp/src/llama.cpp +14 -0
  194. package/cpp/rn-completion.cpp +60 -5
  195. package/ios/include/chat.h +73 -6
  196. package/ios/include/common/minja/chat-template.hpp +9 -5
  197. package/ios/include/common/minja/minja.hpp +69 -36
  198. package/ios/include/common.h +21 -17
  199. package/ios/include/llama.h +62 -125
  200. package/ios/libs/llama.xcframework/Info.plist +19 -19
  201. package/ios/libs/llama.xcframework/ios-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  202. package/ios/libs/llama.xcframework/ios-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4617 -4487
  203. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/ggml-opt.h +237 -0
  204. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/ggml.h +26 -7
  205. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/Headers/llama.h +62 -125
  206. package/ios/libs/llama.xcframework/ios-arm64/llama.framework/llama +0 -0
  207. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  208. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4638 -4508
  209. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3557 -3435
  210. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/ggml-opt.h +237 -0
  211. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +26 -7
  212. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/Headers/llama.h +62 -125
  213. package/ios/libs/llama.xcframework/ios-arm64_x86_64-simulator/llama.framework/llama +0 -0
  214. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  215. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4638 -4508
  216. package/ios/libs/llama.xcframework/macos-arm64_x86_64/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3559 -3437
  217. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/ggml-opt.h +237 -0
  218. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/ggml.h +26 -7
  219. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Headers/llama.h +62 -125
  220. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/ggml-opt.h +237 -0
  221. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/ggml.h +26 -7
  222. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/Headers/llama.h +62 -125
  223. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/A/llama +0 -0
  224. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/ggml-opt.h +237 -0
  225. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/ggml.h +26 -7
  226. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/Headers/llama.h +62 -125
  227. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/Versions/Current/llama +0 -0
  228. package/ios/libs/llama.xcframework/macos-arm64_x86_64/llama.framework/llama +0 -0
  229. package/ios/libs/llama.xcframework/tvos-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  230. package/ios/libs/llama.xcframework/tvos-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4616 -4487
  231. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/ggml-opt.h +237 -0
  232. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/ggml.h +26 -7
  233. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/Headers/llama.h +62 -125
  234. package/ios/libs/llama.xcframework/tvos-arm64/llama.framework/llama +0 -0
  235. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  236. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4637 -4508
  237. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3556 -3435
  238. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/ggml-opt.h +237 -0
  239. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +26 -7
  240. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/Headers/llama.h +62 -125
  241. package/ios/libs/llama.xcframework/tvos-arm64_x86_64-simulator/llama.framework/llama +0 -0
  242. package/ios/libs/llama.xcframework/xros-arm64/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  243. package/ios/libs/llama.xcframework/xros-arm64/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4653 -4523
  244. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/ggml-opt.h +237 -0
  245. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/ggml.h +26 -7
  246. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/Headers/llama.h +62 -125
  247. package/ios/libs/llama.xcframework/xros-arm64/llama.framework/llama +0 -0
  248. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/DWARF/llama +0 -0
  249. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/aarch64/llama.yml +4674 -4544
  250. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/dSYMs/llama.dSYM/Contents/Resources/Relocations/x86_64/llama.yml +3587 -3465
  251. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/ggml-opt.h +237 -0
  252. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/ggml.h +26 -7
  253. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/Headers/llama.h +62 -125
  254. package/ios/libs/llama.xcframework/xros-arm64_x86_64-simulator/llama.framework/llama +0 -0
  255. package/package.json +1 -1
  256. package/cpp/llama.cpp/common/stb_image.h +0 -7988
  257. package/cpp/llama.cpp/models/ggml-vocab-chameleon.gguf.inp +0 -112
  258. package/cpp/llama.cpp/models/ggml-vocab-chameleon.gguf.out +0 -46
  259. package/cpp/llama.cpp/models/ggml-vocab-deepseek-r1-qwen.gguf.inp +0 -112
  260. package/cpp/llama.cpp/models/ggml-vocab-deepseek-r1-qwen.gguf.out +0 -46
  261. package/cpp/llama.cpp/models/ggml-vocab-gpt-4o.gguf.inp +0 -112
  262. package/cpp/llama.cpp/models/ggml-vocab-gpt-4o.gguf.out +0 -46
  263. package/cpp/llama.cpp/models/ggml-vocab-llama4.gguf.inp +0 -112
  264. package/cpp/llama.cpp/models/ggml-vocab-llama4.gguf.out +0 -46
  265. package/cpp/llama.cpp/models/ggml-vocab-pixtral.gguf.inp +0 -112
  266. package/cpp/llama.cpp/models/ggml-vocab-pixtral.gguf.out +0 -46
  267. package/cpp/llama.cpp/models/ggml-vocab-roberta-bpe.gguf.inp +0 -112
  268. package/cpp/llama.cpp/models/ggml-vocab-roberta-bpe.gguf.out +0 -46
package/cpp/llama.cpp/requirements/requirements-convert_hf_to_gguf_update.txt CHANGED
@@ -1,3 +1,7 @@
1
1
  -r ./requirements-convert_legacy_llama.txt
2
2
  --extra-index-url https://download.pytorch.org/whl/cpu
3
- torch~=2.2.1
3
+ torch~=2.2.1; platform_machine != "s390x"
4
+
5
+ # torch s390x packages can only be found from nightly builds
6
+ --extra-index-url https://download.pytorch.org/whl/nightly
7
+ torch>=0.0.0.dev0; platform_machine == "s390x"
package/cpp/llama.cpp/requirements/requirements-convert_lora_to_gguf.txt CHANGED
@@ -1,2 +1,4 @@
1
1
  -r ./requirements-convert_hf_to_gguf.txt
2
2
  --extra-index-url https://download.pytorch.org/whl/cpu
3
+ # torch s390x packages can only be found from nightly builds
4
+ --extra-index-url https://download.pytorch.org/whl/nightly
package/cpp/llama.cpp/requirements/requirements-gguf_editor_gui.txt CHANGED
@@ -1,3 +1,3 @@
1
1
  numpy~=1.26.4
2
2
  PySide6~=6.9.0
3
- gguf>=0.16.0
3
+ gguf>=0.17.0
package/cpp/llama.cpp/src/CMakeLists.txt CHANGED
@@ -14,6 +14,7 @@ add_library(llama
14
14
  llama-batch.cpp
15
15
  llama-chat.cpp
16
16
  llama-context.cpp
17
+ llama-cparams.cpp
17
18
  llama-grammar.cpp
18
19
  llama-graph.cpp
19
20
  llama-hparams.cpp
@@ -23,6 +24,7 @@ add_library(llama
23
24
  llama-memory.cpp
24
25
  llama-mmap.cpp
25
26
  llama-model-loader.cpp
27
+ llama-model-saver.cpp
26
28
  llama-model.cpp
27
29
  llama-quant.cpp
28
30
  llama-sampling.cpp
package/cpp/llama.cpp/src/llama-arch.cpp CHANGED
@@ -174,6 +174,8 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
174
174
  { LLM_KV_CONVNEXT_EMBEDDING_LENGTH, "%s.convnext.embedding_length" },
175
175
  { LLM_KV_CONVNEXT_BLOCK_COUNT, "%s.convnext.block_count" },
176
176
 
177
+ { LLM_KV_CLASSIFIER_OUTPUT_LABELS, "%s.classifier.output_labels" },
178
+
177
179
  { LLM_KV_TOKENIZER_MODEL, "tokenizer.ggml.model" },
178
180
  { LLM_KV_TOKENIZER_PRE, "tokenizer.ggml.pre" },
179
181
  { LLM_KV_TOKENIZER_LIST, "tokenizer.ggml.tokens" },
@@ -448,6 +450,7 @@ static const std::map<llm_arch, std::map<llm_tensor, const char *>> LLM_TENSOR_N
448
450
  { LLM_TENSOR_TOKEN_TYPES, "token_types" },
449
451
  { LLM_TENSOR_POS_EMBD, "position_embd" },
450
452
  { LLM_TENSOR_ATTN_OUT_NORM, "blk.%d.attn_output_norm" },
453
+ { LLM_TENSOR_ATTN_QKV, "blk.%d.attn_qkv" },
451
454
  { LLM_TENSOR_ATTN_Q, "blk.%d.attn_q" },
452
455
  { LLM_TENSOR_ATTN_K, "blk.%d.attn_k" },
453
456
  { LLM_TENSOR_ATTN_V, "blk.%d.attn_v" },
@@ -1481,6 +1484,9 @@ static const std::map<llm_arch, std::map<llm_tensor, const char *>> LLM_TENSOR_N
1481
1484
  { LLM_TENSOR_FFN_GATE_EXPS, "blk.%d.ffn_gate_exps" },
1482
1485
  { LLM_TENSOR_FFN_DOWN_EXPS, "blk.%d.ffn_down_exps" },
1483
1486
  { LLM_TENSOR_FFN_UP_EXPS, "blk.%d.ffn_up_exps" },
1487
+ { LLM_TENSOR_FFN_GATE_SHEXP, "blk.%d.ffn_gate_shexp" },
1488
+ { LLM_TENSOR_FFN_DOWN_SHEXP, "blk.%d.ffn_down_shexp" },
1489
+ { LLM_TENSOR_FFN_UP_SHEXP, "blk.%d.ffn_up_shexp" },
1484
1490
  },
1485
1491
  },
1486
1492
  {
package/cpp/llama.cpp/src/llama-arch.h CHANGED
@@ -213,6 +213,8 @@ enum llm_kv {
213
213
  LLM_KV_CONVNEXT_EMBEDDING_LENGTH,
214
214
  LLM_KV_CONVNEXT_BLOCK_COUNT,
215
215
 
216
+ LLM_KV_CLASSIFIER_OUTPUT_LABELS,
217
+
216
218
  // deprecated:
217
219
  LLM_KV_TOKENIZER_PREFIX_ID,
218
220
  LLM_KV_TOKENIZER_SUFFIX_ID,
package/cpp/llama.cpp/src/llama-batch.cpp CHANGED
@@ -1,5 +1,6 @@
1
1
  #include "llama-batch.h"
2
2
 
3
+ #include <cassert>
3
4
  #include <cstring>
4
5
  #include <algorithm>
5
6
 
@@ -281,9 +282,10 @@ llama_batch_allocr::llama_batch_allocr(struct llama_batch in_batch, llama_pos p0
281
282
  batch = in_batch;
282
283
  GGML_ASSERT(batch.n_tokens > 0);
283
284
  if (!batch.pos) {
285
+ assert(p0 >= 0);
284
286
  pos.resize(batch.n_tokens);
285
287
  for (int32_t i = 0; i < batch.n_tokens; i++) {
286
- pos[i] = i + p0;
288
+ pos[i] = p0 + i;
287
289
  }
288
290
  batch.pos = pos.data();
289
291
  }