cui-llama.rn 1.3.0 → 1.3.4

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 (77) hide show
  1. package/android/src/main/CMakeLists.txt +9 -6
  2. package/android/src/main/java/com/rnllama/LlamaContext.java +4 -4
  3. package/android/src/main/jni.cpp +15 -15
  4. package/cpp/common.cpp +1962 -1682
  5. package/cpp/common.h +645 -600
  6. package/cpp/ggml-alloc.c +1038 -1040
  7. package/cpp/ggml-alloc.h +76 -76
  8. package/cpp/ggml-backend-impl.h +256 -216
  9. package/cpp/ggml-backend-reg.cpp +552 -195
  10. package/cpp/ggml-backend.cpp +1999 -1997
  11. package/cpp/ggml-backend.h +352 -328
  12. package/cpp/ggml-common.h +1853 -1853
  13. package/cpp/ggml-cpp.h +38 -38
  14. package/cpp/{ggml-cpu-aarch64.c → ggml-cpu-aarch64.cpp} +4262 -3560
  15. package/cpp/ggml-cpu-aarch64.h +8 -30
  16. package/cpp/ggml-cpu-impl.h +386 -371
  17. package/cpp/ggml-cpu-quants.c +10835 -10822
  18. package/cpp/ggml-cpu-quants.h +63 -63
  19. package/cpp/ggml-cpu-traits.cpp +36 -0
  20. package/cpp/ggml-cpu-traits.h +38 -0
  21. package/cpp/ggml-cpu.c +14122 -13975
  22. package/cpp/ggml-cpu.cpp +618 -663
  23. package/cpp/ggml-cpu.h +135 -177
  24. package/cpp/ggml-impl.h +556 -550
  25. package/cpp/ggml-metal.h +66 -66
  26. package/cpp/ggml-metal.m +4884 -4294
  27. package/cpp/ggml-quants.c +5238 -5247
  28. package/cpp/ggml-quants.h +100 -100
  29. package/cpp/ggml-threading.cpp +12 -12
  30. package/cpp/ggml-threading.h +14 -12
  31. package/cpp/ggml.c +7707 -8180
  32. package/cpp/ggml.h +2286 -2411
  33. package/cpp/json-schema-to-grammar.cpp +1045 -0
  34. package/cpp/json-schema-to-grammar.h +8 -0
  35. package/cpp/json.hpp +24766 -0
  36. package/cpp/llama-grammar.cpp +1138 -1138
  37. package/cpp/llama-grammar.h +144 -144
  38. package/cpp/llama-impl.h +181 -181
  39. package/cpp/llama-sampling.cpp +2293 -2348
  40. package/cpp/llama-sampling.h +48 -48
  41. package/cpp/llama-vocab.cpp +1985 -1984
  42. package/cpp/llama-vocab.h +170 -170
  43. package/cpp/llama.cpp +22836 -22132
  44. package/cpp/llama.h +1263 -1253
  45. package/cpp/log.cpp +401 -401
  46. package/cpp/log.h +121 -121
  47. package/cpp/rn-llama.hpp +6 -6
  48. package/cpp/sampling.cpp +500 -466
  49. package/cpp/sampling.h +22 -1
  50. package/cpp/sgemm.cpp +1884 -1884
  51. package/cpp/speculative.cpp +274 -0
  52. package/cpp/speculative.h +28 -0
  53. package/cpp/unicode.cpp +62 -51
  54. package/cpp/unicode.h +9 -10
  55. package/ios/RNLlamaContext.mm +13 -0
  56. package/lib/commonjs/NativeRNLlama.js.map +1 -1
  57. package/lib/commonjs/grammar.js +4 -2
  58. package/lib/commonjs/grammar.js.map +1 -1
  59. package/lib/commonjs/index.js +38 -1
  60. package/lib/commonjs/index.js.map +1 -1
  61. package/lib/module/NativeRNLlama.js.map +1 -1
  62. package/lib/module/grammar.js +2 -1
  63. package/lib/module/grammar.js.map +1 -1
  64. package/lib/module/index.js +36 -0
  65. package/lib/module/index.js.map +1 -1
  66. package/lib/typescript/NativeRNLlama.d.ts +95 -6
  67. package/lib/typescript/NativeRNLlama.d.ts.map +1 -1
  68. package/lib/typescript/grammar.d.ts +5 -6
  69. package/lib/typescript/grammar.d.ts.map +1 -1
  70. package/lib/typescript/index.d.ts +40 -4
  71. package/lib/typescript/index.d.ts.map +1 -1
  72. package/package.json +2 -1
  73. package/src/NativeRNLlama.ts +99 -12
  74. package/src/grammar.ts +10 -8
  75. package/src/index.ts +68 -3
  76. package/cpp/ggml-aarch64.c +0 -129
  77. package/cpp/ggml-aarch64.h +0 -19
package/cpp/llama-sampling.h CHANGED
@@ -1,48 +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);
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);
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);