@fugood/llama.node 1.4.1 → 1.4.3

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 (42) hide show
  1. package/CMakeLists.txt +1 -1
  2. package/lib/binding.js +3 -0
  3. package/lib/binding.ts +2 -0
  4. package/package.json +16 -16
  5. package/scripts/llama.cpp.patch +25 -11
  6. package/src/LlamaContext.cpp +2 -2
  7. package/src/llama.cpp/CMakeLists.txt +21 -6
  8. package/src/llama.cpp/common/CMakeLists.txt +6 -0
  9. package/src/llama.cpp/common/arg.cpp +65 -16
  10. package/src/llama.cpp/common/chat-parser.cpp +40 -0
  11. package/src/llama.cpp/common/chat-peg-parser.cpp +110 -0
  12. package/src/llama.cpp/common/chat-peg-parser.h +105 -0
  13. package/src/llama.cpp/common/chat.cpp +40 -29
  14. package/src/llama.cpp/common/chat.h +10 -1
  15. package/src/llama.cpp/common/common.cpp +24 -5
  16. package/src/llama.cpp/common/common.h +16 -5
  17. package/src/llama.cpp/common/download.cpp +18 -8
  18. package/src/llama.cpp/common/download.h +3 -1
  19. package/src/llama.cpp/common/json-schema-to-grammar.cpp +1 -1
  20. package/src/llama.cpp/common/log.cpp +15 -1
  21. package/src/llama.cpp/common/log.h +19 -12
  22. package/src/llama.cpp/common/peg-parser.cpp +1712 -0
  23. package/src/llama.cpp/common/peg-parser.h +459 -0
  24. package/src/llama.cpp/common/unicode.cpp +64 -0
  25. package/src/llama.cpp/common/unicode.h +22 -0
  26. package/src/llama.cpp/ggml/CMakeLists.txt +48 -48
  27. package/src/llama.cpp/ggml/include/ggml.h +7 -2
  28. package/src/llama.cpp/ggml/src/CMakeLists.txt +0 -4
  29. package/src/llama.cpp/ggml/src/ggml-cpu/arch/arm/cpu-feats.cpp +4 -0
  30. package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c +10 -13
  31. package/src/llama.cpp/ggml/src/ggml-cpu/ops.cpp +60 -1
  32. package/src/llama.cpp/src/CMakeLists.txt +1 -0
  33. package/src/llama.cpp/src/llama-arch.cpp +30 -1
  34. package/src/llama.cpp/src/llama-arch.h +3 -0
  35. package/src/llama.cpp/src/llama-graph.cpp +3 -6
  36. package/src/llama.cpp/src/llama-hparams.h +2 -2
  37. package/src/llama.cpp/src/llama-impl.h +1 -1
  38. package/src/llama.cpp/src/llama-mmap.cpp +1 -1
  39. package/src/llama.cpp/src/llama-model.cpp +50 -6
  40. package/src/llama.cpp/src/llama-vocab.cpp +1 -2
  41. package/src/llama.cpp/src/models/mistral3.cpp +160 -0
  42. package/src/llama.cpp/src/models/models.h +4 -0
package/src/llama.cpp/common/log.h CHANGED
@@ -21,8 +21,14 @@
21
21
  # define LOG_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
22
22
  #endif
23
23
 
24
- #define LOG_DEFAULT_DEBUG 1
25
- #define LOG_DEFAULT_LLAMA 0
24
+ #define LOG_LEVEL_DEBUG 4
25
+ #define LOG_LEVEL_INFO 3
26
+ #define LOG_LEVEL_WARN 2
27
+ #define LOG_LEVEL_ERROR 1
28
+ #define LOG_LEVEL_OUTPUT 0 // output data from tools
29
+
30
+ #define LOG_DEFAULT_DEBUG LOG_LEVEL_DEBUG
31
+ #define LOG_DEFAULT_LLAMA LOG_LEVEL_INFO
26
32
 
27
33
  enum log_colors {
28
34
  LOG_COLORS_AUTO = -1,
@@ -67,10 +73,11 @@ void common_log_add(struct common_log * log, enum ggml_log_level level, const ch
67
73
  // 0.00.090.578 I llm_load_tensors: offloading 32 repeating layers to GPU
68
74
  // 0.00.090.579 I llm_load_tensors: offloading non-repeating layers to GPU
69
75
  //
70
- // I - info (stdout, V = 0)
71
- // W - warning (stderr, V = 0)
72
- // E - error (stderr, V = 0)
73
76
  // D - debug (stderr, V = LOG_DEFAULT_DEBUG)
77
+ // I - info (stdout, V = LOG_DEFAULT_INFO)
78
+ // W - warning (stderr, V = LOG_DEFAULT_WARN)
79
+ // E - error (stderr, V = LOG_DEFAULT_ERROR)
80
+ // O - output (stdout, V = LOG_DEFAULT_OUTPUT)
74
81
  //
75
82
 
76
83
  void common_log_set_file (struct common_log * log, const char * file); // not thread-safe
@@ -95,14 +102,14 @@ void common_log_set_timestamps(struct common_log * log, bool timestamps); // w
95
102
  } \
96
103
  } while (0)
97
104
 
98
- #define LOG(...) LOG_TMPL(GGML_LOG_LEVEL_NONE, 0, __VA_ARGS__)
99
- #define LOGV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_NONE, verbosity, __VA_ARGS__)
105
+ #define LOG(...) LOG_TMPL(GGML_LOG_LEVEL_NONE, LOG_LEVEL_OUTPUT, __VA_ARGS__)
106
+ #define LOGV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_NONE, verbosity, __VA_ARGS__)
100
107
 
101
- #define LOG_INF(...) LOG_TMPL(GGML_LOG_LEVEL_INFO, 0, __VA_ARGS__)
102
- #define LOG_WRN(...) LOG_TMPL(GGML_LOG_LEVEL_WARN, 0, __VA_ARGS__)
103
- #define LOG_ERR(...) LOG_TMPL(GGML_LOG_LEVEL_ERROR, 0, __VA_ARGS__)
104
- #define LOG_DBG(...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, LOG_DEFAULT_DEBUG, __VA_ARGS__)
105
- #define LOG_CNT(...) LOG_TMPL(GGML_LOG_LEVEL_CONT, 0, __VA_ARGS__)
108
+ #define LOG_DBG(...) LOG_TMPL(GGML_LOG_LEVEL_DEBUG, LOG_LEVEL_DEBUG, __VA_ARGS__)
109
+ #define LOG_INF(...) LOG_TMPL(GGML_LOG_LEVEL_INFO, LOG_LEVEL_INFO, __VA_ARGS__)
110
+ #define LOG_WRN(...) LOG_TMPL(GGML_LOG_LEVEL_WARN, LOG_LEVEL_WARN, __VA_ARGS__)
111
+ #define LOG_ERR(...) LOG_TMPL(GGML_LOG_LEVEL_ERROR, LOG_LEVEL_ERROR, __VA_ARGS__)
112
+ #define LOG_CNT(...) LOG_TMPL(GGML_LOG_LEVEL_CONT, LOG_LEVEL_INFO, __VA_ARGS__) // same as INFO
106
113
 
107
114
  #define LOG_INFV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_INFO, verbosity, __VA_ARGS__)
108
115
  #define LOG_WRNV(verbosity, ...) LOG_TMPL(GGML_LOG_LEVEL_WARN, verbosity, __VA_ARGS__)