cui-llama.rn 1.2.4 → 1.3.0

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 (70) hide show
  1. package/README.md +3 -4
  2. package/android/src/main/CMakeLists.txt +21 -5
  3. package/android/src/main/java/com/rnllama/LlamaContext.java +115 -30
  4. package/android/src/main/java/com/rnllama/RNLlama.java +40 -7
  5. package/android/src/main/jni.cpp +222 -36
  6. package/android/src/newarch/java/com/rnllama/RNLlamaModule.java +9 -4
  7. package/android/src/oldarch/java/com/rnllama/RNLlamaModule.java +9 -4
  8. package/cpp/common.cpp +1682 -2122
  9. package/cpp/common.h +600 -594
  10. package/cpp/ggml-aarch64.c +129 -3209
  11. package/cpp/ggml-aarch64.h +19 -39
  12. package/cpp/ggml-alloc.c +1040 -1040
  13. package/cpp/ggml-alloc.h +76 -76
  14. package/cpp/ggml-backend-impl.h +216 -227
  15. package/cpp/ggml-backend-reg.cpp +195 -0
  16. package/cpp/ggml-backend.cpp +1997 -2625
  17. package/cpp/ggml-backend.h +328 -326
  18. package/cpp/ggml-common.h +1853 -1853
  19. package/cpp/ggml-cpp.h +38 -0
  20. package/cpp/ggml-cpu-aarch64.c +3560 -0
  21. package/cpp/ggml-cpu-aarch64.h +30 -0
  22. package/cpp/ggml-cpu-impl.h +371 -614
  23. package/cpp/ggml-cpu-quants.c +10822 -0
  24. package/cpp/ggml-cpu-quants.h +63 -0
  25. package/cpp/ggml-cpu.c +13975 -0
  26. package/cpp/ggml-cpu.cpp +663 -0
  27. package/cpp/ggml-cpu.h +177 -0
  28. package/cpp/ggml-impl.h +550 -209
  29. package/cpp/ggml-metal.h +66 -66
  30. package/cpp/ggml-metal.m +4294 -3819
  31. package/cpp/ggml-quants.c +5247 -15752
  32. package/cpp/ggml-quants.h +100 -147
  33. package/cpp/ggml-threading.cpp +12 -0
  34. package/cpp/ggml-threading.h +12 -0
  35. package/cpp/ggml.c +8180 -23464
  36. package/cpp/ggml.h +2411 -2562
  37. package/cpp/llama-grammar.cpp +1138 -1138
  38. package/cpp/llama-grammar.h +144 -144
  39. package/cpp/llama-impl.h +181 -181
  40. package/cpp/llama-sampling.cpp +2348 -2194
  41. package/cpp/llama-sampling.h +48 -30
  42. package/cpp/llama-vocab.cpp +1984 -1968
  43. package/cpp/llama-vocab.h +170 -165
  44. package/cpp/llama.cpp +22132 -21969
  45. package/cpp/llama.h +1253 -1253
  46. package/cpp/log.cpp +401 -401
  47. package/cpp/log.h +121 -121
  48. package/cpp/rn-llama.hpp +83 -19
  49. package/cpp/sampling.cpp +466 -458
  50. package/cpp/sgemm.cpp +1884 -1219
  51. package/ios/RNLlama.mm +43 -20
  52. package/ios/RNLlamaContext.h +9 -3
  53. package/ios/RNLlamaContext.mm +133 -33
  54. package/jest/mock.js +0 -1
  55. package/lib/commonjs/NativeRNLlama.js.map +1 -1
  56. package/lib/commonjs/index.js +52 -15
  57. package/lib/commonjs/index.js.map +1 -1
  58. package/lib/module/NativeRNLlama.js.map +1 -1
  59. package/lib/module/index.js +51 -15
  60. package/lib/module/index.js.map +1 -1
  61. package/lib/typescript/NativeRNLlama.d.ts +29 -6
  62. package/lib/typescript/NativeRNLlama.d.ts.map +1 -1
  63. package/lib/typescript/index.d.ts +12 -5
  64. package/lib/typescript/index.d.ts.map +1 -1
  65. package/package.json +1 -1
  66. package/src/NativeRNLlama.ts +41 -7
  67. package/src/index.ts +82 -27
  68. package/cpp/json-schema-to-grammar.cpp +0 -1045
  69. package/cpp/json-schema-to-grammar.h +0 -8
  70. package/cpp/json.hpp +0 -24766
package/cpp/llama-sampling.h CHANGED
@@ -1,30 +1,48 @@
1
- #pragma once
2
-
3
- // TODO: rename llama-sampling.h/.cpp to llama-sampler.h/.cpp ?
4
-
5
- #include "llama-grammar.h"
6
-
7
- struct llama_vocab;
8
- struct llama_grammar;
9
-
10
- // sampler chain
11
-
12
- struct llama_sampler_chain {
13
- llama_sampler_chain_params params;
14
-
15
- std::vector<struct llama_sampler *> samplers;
16
-
17
- // timing
18
-
19
- mutable int64_t t_sample_us;
20
-
21
- mutable int32_t n_sample;
22
- };
23
-
24
- struct llama_sampler * llama_sampler_init_grammar_impl(
25
- const struct llama_vocab & vocab,
26
- const char * grammar_str,
27
- const char * grammar_root);
28
-
29
- struct llama_sampler * llama_sampler_init_infill_impl(
30
- const struct llama_vocab & vocab);
1
+ #pragma once
2
+
3
+ // TODO: rename llama-sampling.h/.cpp to llama-sampler.h/.cpp ?
4
+
5
+ #include "llama-grammar.h"
6
+
7
+ struct llama_vocab;
8
+ struct llama_grammar;
9
+
10
+ // sampler chain
11
+
12
+ struct llama_sampler_chain {
13
+ llama_sampler_chain_params params;
14
+
15
+ std::vector<struct llama_sampler *> samplers;
16
+
17
+ // timing
18
+
19
+ mutable int64_t t_sample_us;
20
+
21
+ mutable int32_t n_sample;
22
+ };
23
+
24
+ struct llama_sampler * llama_sampler_init_grammar_impl(
25
+ const struct llama_vocab & vocab,
26
+ const char * grammar_str,
27
+ const char * grammar_root);
28
+
29
+ struct llama_sampler * llama_sampler_init_infill_impl(
30
+ const struct llama_vocab & vocab);
31
+
32
+ struct llama_sampler * llama_sampler_init_dry_impl(
33
+ const struct llama_vocab & vocab,
34
+ int32_t context_size,
35
+ float dry_multiplier,
36
+ float dry_base,
37
+ int32_t dry_allowed_length,
38
+ int32_t dry_penalty_last_n,
39
+ const char ** seq_breakers,
40
+ size_t num_breakers);
41
+
42
+ struct llama_sampler * llama_sampler_init_dry_testing(
43
+ int32_t context_size,
44
+ float dry_multiplier,
45
+ float dry_base,
46
+ int32_t dry_allowed_length,
47
+ int32_t dry_penalty_last_n,
48
+ const std::vector<std::vector<llama_token>>& seq_breakers);