Repository navigation
Expand file tree
/
Copy pathllama_cpp.py
More file actions
1188 lines (909 loc) · 39.7 KB
/
Copy pathllama_cpp.py
File metadata and controls
1188 lines (909 loc) · 39.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
import sys
import os
import ctypes
from ctypes import (
c_double,
c_int,
c_float,
c_char_p,
c_int32,
c_uint32,
c_void_p,
c_bool,
POINTER,
_Pointer, # type: ignore
Structure,
Array,
c_uint8,
c_size_t,
)
import pathlib
from typing import List, Union
# Load the library
def _load_shared_library(lib_base_name: str):
# Construct the paths to the possible shared library names
_base_path = pathlib.Path(__file__).parent.resolve()
# Searching for the library in the current directory under the name "libllama" (default name
# for llamacpp) and "llama" (default name for this repo)
_lib_paths: List[pathlib.Path] = []
# Determine the file extension based on the platform
if sys.platform.startswith("linux"):
_lib_paths += [
_base_path / f"lib{lib_base_name}.so",
]
elif sys.platform == "darwin":
_lib_paths += [
_base_path / f"lib{lib_base_name}.so",
_base_path / f"lib{lib_base_name}.dylib",
]
elif sys.platform == "win32":
_lib_paths += [
_base_path / f"{lib_base_name}.dll",
]
else:
raise RuntimeError("Unsupported platform")
if "LLAMA_CPP_LIB" in os.environ:
lib_base_name = os.environ["LLAMA_CPP_LIB"]
_lib = pathlib.Path(lib_base_name)
_base_path = _lib.parent.resolve()
_lib_paths = [_lib.resolve()]
cdll_args = dict() # type: ignore
# Add the library directory to the DLL search path on Windows (if needed)
if sys.platform == "win32" and sys.version_info >= (3, 8):
os.add_dll_directory(str(_base_path))
if "CUDA_PATH" in os.environ:
os.add_dll_directory(os.path.join(os.environ["CUDA_PATH"], "bin"))
os.add_dll_directory(os.path.join(os.environ["CUDA_PATH"], "lib"))
cdll_args["winmode"] = 0
# Try to load the shared library, handling potential errors
for _lib_path in _lib_paths:
if _lib_path.exists():
try:
return ctypes.CDLL(str(_lib_path), **cdll_args)
except Exception as e:
raise RuntimeError(f"Failed to load shared library '{_lib_path}': {e}")
raise FileNotFoundError(
f"Shared library with base name '{lib_base_name}' not found"
)
# Specify the base name of the shared library to load
_lib_base_name = "llama"
# Load the library
_lib = _load_shared_library(_lib_base_name)
# Misc
c_float_p = POINTER(c_float)
c_uint8_p = POINTER(c_uint8)
c_size_t_p = POINTER(c_size_t)
# llama.h bindings
GGML_USE_CUBLAS = hasattr(_lib, "ggml_init_cublas")
GGML_CUDA_MAX_DEVICES = ctypes.c_int(16)
LLAMA_MAX_DEVICES = GGML_CUDA_MAX_DEVICES if GGML_USE_CUBLAS else ctypes.c_int(1)
# #define LLAMA_FILE_MAGIC_GGJT 0x67676a74u // 'ggjt'
LLAMA_FILE_MAGIC_GGJT = ctypes.c_uint(0x67676A74)
# #define LLAMA_FILE_MAGIC_GGLA 0x67676c61u // 'ggla'
LLAMA_FILE_MAGIC_GGLA = ctypes.c_uint(0x67676C61)
# #define LLAMA_FILE_MAGIC_GGMF 0x67676d66u // 'ggmf'
LLAMA_FILE_MAGIC_GGMF = ctypes.c_uint(0x67676D66)
# #define LLAMA_FILE_MAGIC_GGML 0x67676d6cu // 'ggml'
LLAMA_FILE_MAGIC_GGML = ctypes.c_uint(0x67676D6C)
# #define LLAMA_FILE_MAGIC_GGSN 0x6767736eu // 'ggsn'
LLAMA_FILE_MAGIC_GGSN = ctypes.c_uint(0x6767736E)
# #define LLAMA_FILE_VERSION 3
LLAMA_FILE_VERSION = c_int(3)
LLAMA_FILE_MAGIC = LLAMA_FILE_MAGIC_GGJT
LLAMA_FILE_MAGIC_UNVERSIONED = LLAMA_FILE_MAGIC_GGML
LLAMA_SESSION_MAGIC = LLAMA_FILE_MAGIC_GGSN
LLAMA_SESSION_VERSION = c_int(1)
# #define LLAMA_DEFAULT_SEED 0xFFFFFFFF
LLAMA_DEFAULT_SEED = c_int(0xFFFFFFFF)
# struct llama_model;
llama_model_p = c_void_p
# struct llama_context;
llama_context_p = c_void_p
# typedef int llama_token;
llama_token = c_int
llama_token_p = POINTER(llama_token)
# typedef struct llama_token_data {
# llama_token id; // token id
# float logit; // log-odds of the token
# float p; // probability of the token
# } llama_token_data;
class llama_token_data(Structure):
_fields_ = [
("id", llama_token),
("logit", c_float),
("p", c_float),
]
llama_token_data_p = POINTER(llama_token_data)
# typedef struct llama_token_data_array {
# llama_token_data * data;
# size_t size;
# bool sorted;
# } llama_token_data_array;
class llama_token_data_array(Structure):
_fields_ = [
("data", llama_token_data_p),
("size", c_size_t),
("sorted", c_bool),
]
llama_token_data_array_p = POINTER(llama_token_data_array)
# typedef void (*llama_progress_callback)(float progress, void *ctx);
llama_progress_callback = ctypes.CFUNCTYPE(None, c_float, c_void_p)
# struct llama_context_params {
# uint32_t seed; // RNG seed, -1 for random
# int32_t n_ctx; // text context
# int32_t n_batch; // prompt processing batch size
# int32_t n_gpu_layers; // number of layers to store in VRAM
# int32_t main_gpu; // the GPU that is used for scratch and small tensors
# float tensor_split[LLAMA_MAX_DEVICES]; // how to split layers across multiple GPUs
# // ref: https://github.com/ggerganov/llama.cpp/pull/2054
# float rope_freq_base; // RoPE base frequency
# float rope_freq_scale; // RoPE frequency scaling factor
# // called with a progress value between 0 and 1, pass NULL to disable
# llama_progress_callback progress_callback;
# // context pointer passed to the progress callback
# void * progress_callback_user_data;
# // Keep the booleans together to avoid misalignment during copy-by-value.
# bool low_vram; // if true, reduce VRAM usage at the cost of performance
# bool f16_kv; // use fp16 for KV cache
# bool logits_all; // the llama_eval() call computes all logits, not just the last one
# bool vocab_only; // only load the vocabulary, no weights
# bool use_mmap; // use mmap if possible
# bool use_mlock; // force system to keep model in RAM
# bool embedding; // embedding mode only
# };
class llama_context_params(Structure):
_fields_ = [
("seed", c_uint32),
("n_ctx", c_int32),
("rope_freq_base", c_float),
("rope_freq_scale", c_float),
("n_batch", c_int32),
("n_gpu_layers", c_int32),
("main_gpu", c_int32),
("tensor_split", c_float * LLAMA_MAX_DEVICES.value),
("rope_freq_base", c_float),
("rope_freq_scale", c_float),
("progress_callback", llama_progress_callback),
("progress_callback_user_data", c_void_p),
("low_vram", c_bool),
("f16_kv", c_bool),
("logits_all", c_bool),
("vocab_only", c_bool),
("use_mmap", c_bool),
("use_mlock", c_bool),
("embedding", c_bool),
]
llama_context_params_p = POINTER(llama_context_params)
# enum llama_ftype {
# LLAMA_FTYPE_ALL_F32 = 0,
# LLAMA_FTYPE_MOSTLY_F16 = 1, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q4_0 = 2, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q4_1 = 3, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
# // LLAMA_FTYPE_MOSTLY_Q4_2 = 5, // support has been removed
# // LLAMA_FTYPE_MOSTLY_Q4_3 = 6, // support has been removed
# LLAMA_FTYPE_MOSTLY_Q8_0 = 7, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q5_0 = 8, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q5_1 = 9, // except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q2_K = 10,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q3_K_S = 11,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q3_K_M = 12,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q3_K_L = 13,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q4_K_S = 14,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q4_K_M = 15,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q5_K_S = 16,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q5_K_M = 17,// except 1d tensors
# LLAMA_FTYPE_MOSTLY_Q6_K = 18,// except 1d tensors
# };
LLAMA_FTYPE_ALL_F32 = c_int(0)
LLAMA_FTYPE_MOSTLY_F16 = c_int(1)
LLAMA_FTYPE_MOSTLY_Q4_0 = c_int(2)
LLAMA_FTYPE_MOSTLY_Q4_1 = c_int(3)
LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = c_int(4)
LLAMA_FTYPE_MOSTLY_Q8_0 = c_int(7)
LLAMA_FTYPE_MOSTLY_Q5_0 = c_int(8)
LLAMA_FTYPE_MOSTLY_Q5_1 = c_int(9)
LLAMA_FTYPE_MOSTLY_Q2_K = c_int(10)
LLAMA_FTYPE_MOSTLY_Q3_K_S = c_int(11)
LLAMA_FTYPE_MOSTLY_Q3_K_M = c_int(12)
LLAMA_FTYPE_MOSTLY_Q3_K_L = c_int(13)
LLAMA_FTYPE_MOSTLY_Q4_K_S = c_int(14)
LLAMA_FTYPE_MOSTLY_Q4_K_M = c_int(15)
LLAMA_FTYPE_MOSTLY_Q5_K_S = c_int(16)
LLAMA_FTYPE_MOSTLY_Q5_K_M = c_int(17)
LLAMA_FTYPE_MOSTLY_Q6_K = c_int(18)
# // model quantization parameters
# typedef struct llama_model_quantize_params {
# int nthread; // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
# enum llama_ftype ftype; // quantize to this llama_ftype
# bool allow_requantize; // allow quantizing non-f32/f16 tensors
# bool quantize_output_tensor; // quantize output.weight
# } llama_model_quantize_params;
class llama_model_quantize_params(Structure):
_fields_ = [
("nthread", c_int),
("ftype", c_int),
("allow_requantize", c_bool),
("quantize_output_tensor", c_bool),
]
# // performance timing information
# struct llama_timings {
# double t_start_ms;
# double t_end_ms;
# double t_load_ms;
# double t_sample_ms;
# double t_p_eval_ms;
# double t_eval_ms;
# int32_t n_sample;
# int32_t n_p_eval;
# int32_t n_eval;
# };
class llama_timings(Structure):
_fields_ = [
("t_start_ms", c_double),
("t_end_ms", c_double),
("t_load_ms", c_double),
("t_sample_ms", c_double),
("t_p_eval_ms", c_double),
("t_eval_ms", c_double),
("n_sample", c_int32),
("n_p_eval", c_int32),
("n_eval", c_int32),
]
# LLAMA_API struct llama_context_params llama_context_default_params();
def llama_context_default_params() -> llama_context_params:
return _lib.llama_context_default_params()
_lib.llama_context_default_params.argtypes = []
_lib.llama_context_default_params.restype = llama_context_params
# LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
def llama_model_quantize_default_params() -> llama_model_quantize_params:
return _lib.llama_model_quantize_default_params()
_lib.llama_model_quantize_default_params.argtypes = []
_lib.llama_model_quantize_default_params.restype = llama_model_quantize_params
# LLAMA_API bool llama_mmap_supported();
def llama_mmap_supported() -> bool:
return _lib.llama_mmap_supported()
_lib.llama_mmap_supported.argtypes = []
_lib.llama_mmap_supported.restype = c_bool
# LLAMA_API bool llama_mlock_supported();
def llama_mlock_supported() -> bool:
return _lib.llama_mlock_supported()
_lib.llama_mlock_supported.argtypes = []
_lib.llama_mlock_supported.restype = c_bool
# // TODO: not great API - very likely to change
# // Initialize the llama + ggml backend
# // If numa is true, use NUMA optimizations
# // Call once at the start of the program
# LLAMA_API void llama_backend_init(bool numa);
def llama_backend_init(numa: c_bool):
return _lib.llama_backend_init(numa)
_lib.llama_backend_init.argtypes = [c_bool]
_lib.llama_backend_init.restype = None
# // Call once at the end of the program - currently only used for MPI
# LLAMA_API void llama_backend_free();
def llama_backend_free():
return _lib.llama_backend_free()
_lib.llama_backend_free.argtypes = []
_lib.llama_backend_free.restype = None
# LLAMA_API struct llama_model * llama_load_model_from_file(
# const char * path_model,
# struct llama_context_params params);
def llama_load_model_from_file(
path_model: bytes, params: llama_context_params
) -> llama_model_p:
return _lib.llama_load_model_from_file(path_model, params)
_lib.llama_load_model_from_file.argtypes = [c_char_p, llama_context_params]
_lib.llama_load_model_from_file.restype = llama_model_p
# LLAMA_API void llama_free_model(struct llama_model * model);
def llama_free_model(model: llama_model_p):
return _lib.llama_free_model(model)
_lib.llama_free_model.argtypes = [llama_model_p]
_lib.llama_free_model.restype = None
# LLAMA_API struct llama_context * llama_new_context_with_model(
# struct llama_model * model,
# struct llama_context_params params);
def llama_new_context_with_model(
model: llama_model_p, params: llama_context_params
) -> llama_context_p:
return _lib.llama_new_context_with_model(model, params)
_lib.llama_new_context_with_model.argtypes = [llama_model_p, llama_context_params]
_lib.llama_new_context_with_model.restype = llama_context_p
# LLAMA_API int64_t llama_time_us();
def llama_time_us() -> int:
return _lib.llama_time_us()
_lib.llama_time_us.argtypes = []
_lib.llama_time_us.restype = ctypes.c_int64
# // Various functions for loading a ggml llama model.
# // Allocate (almost) all memory needed for the model.
# // Return NULL on failure
# LLAMA_API struct llama_context * llama_init_from_file(
# const char * path_model,
# struct llama_context_params params);
def llama_init_from_file(
path_model: bytes, params: llama_context_params
) -> llama_context_p:
return _lib.llama_init_from_file(path_model, params)
_lib.llama_init_from_file.argtypes = [c_char_p, llama_context_params]
_lib.llama_init_from_file.restype = llama_context_p
# Frees all allocated memory
# LLAMA_API void llama_free(struct llama_context * ctx);
def llama_free(ctx: llama_context_p):
return _lib.llama_free(ctx)
_lib.llama_free.argtypes = [llama_context_p]
_lib.llama_free.restype = None
# // Returns 0 on success
# LLAMA_API int llama_model_quantize(
# const char * fname_inp,
# const char * fname_out,
# const llama_model_quantize_params * params);
def llama_model_quantize(
fname_inp: bytes,
fname_out: bytes,
params, # type: POINTER(llama_model_quantize_params) # type: ignore
) -> int:
return _lib.llama_model_quantize(fname_inp, fname_out, params)
_lib.llama_model_quantize.argtypes = [
c_char_p,
c_char_p,
POINTER(llama_model_quantize_params),
]
_lib.llama_model_quantize.restype = c_int
# Apply a LoRA adapter to a loaded model
# path_base_model is the path to a higher quality model to use as a base for
# the layers modified by the adapter. Can be NULL to use the current loaded model.
# The model needs to be reloaded before applying a new adapter, otherwise the adapter
# will be applied on top of the previous one
# Returns 0 on success
# LLAMA_API int llama_apply_lora_from_file(
# struct llama_context * ctx,
# const char * path_lora,
# const char * path_base_model,
# int n_threads);
def llama_apply_lora_from_file(
ctx: llama_context_p,
path_lora: c_char_p,
path_base_model: c_char_p,
n_threads: c_int,
) -> int:
return _lib.llama_apply_lora_from_file(ctx, path_lora, path_base_model, n_threads)
_lib.llama_apply_lora_from_file.argtypes = [llama_context_p, c_char_p, c_char_p, c_int]
_lib.llama_apply_lora_from_file.restype = c_int
# LLAMA_API int llama_model_apply_lora_from_file(
# const struct llama_model * model,
# const char * path_lora,
# const char * path_base_model,
# int n_threads);
def llama_model_apply_lora_from_file(
model: llama_model_p,
path_lora: Union[c_char_p, bytes],
path_base_model: Union[c_char_p, bytes],
n_threads: c_int,
) -> int:
return _lib.llama_model_apply_lora_from_file(
model, path_lora, path_base_model, n_threads
)
_lib.llama_model_apply_lora_from_file.argtypes = [
llama_model_p,
c_char_p,
c_char_p,
c_int,
]
_lib.llama_model_apply_lora_from_file.restype = c_int
# Returns the number of tokens in the KV cache
# LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
def llama_get_kv_cache_token_count(ctx: llama_context_p) -> int:
return _lib.llama_get_kv_cache_token_count(ctx)
_lib.llama_get_kv_cache_token_count.argtypes = [llama_context_p]
_lib.llama_get_kv_cache_token_count.restype = c_int
# Sets the current rng seed.
# LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, int seed);
def llama_set_rng_seed(ctx: llama_context_p, seed: c_uint32):
return _lib.llama_set_rng_seed(ctx, seed)
_lib.llama_set_rng_seed.argtypes = [llama_context_p, c_int]
_lib.llama_set_rng_seed.restype = None
# Returns the maximum size in bytes of the state (rng, logits, embedding
# and kv_cache) - will often be smaller after compacting tokens
# LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
def llama_get_state_size(ctx: llama_context_p) -> int:
return _lib.llama_get_state_size(ctx)
_lib.llama_get_state_size.argtypes = [llama_context_p]
_lib.llama_get_state_size.restype = c_size_t
# Copies the state to the specified destination address.
# Destination needs to have allocated enough memory.
# Returns the number of bytes copied
# LLAMA_API size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst);
def llama_copy_state_data(
ctx: llama_context_p, dst # type: Array[c_uint8]
) -> int:
return _lib.llama_copy_state_data(ctx, dst)
_lib.llama_copy_state_data.argtypes = [llama_context_p, c_uint8_p]
_lib.llama_copy_state_data.restype = c_size_t
# Set the state reading from the specified address
# Returns the number of bytes read
# LLAMA_API size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src);
def llama_set_state_data(
ctx: llama_context_p, src # type: Array[c_uint8]
) -> int:
return _lib.llama_set_state_data(ctx, src)
_lib.llama_set_state_data.argtypes = [llama_context_p, c_uint8_p]
_lib.llama_set_state_data.restype = c_size_t
# Save/load session file
# LLAMA_API bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out);
def llama_load_session_file(
ctx: llama_context_p,
path_session: bytes,
tokens_out, # type: Array[llama_token]
n_token_capacity: c_size_t,
n_token_count_out, # type: _Pointer[c_size_t]
) -> int:
return _lib.llama_load_session_file(
ctx, path_session, tokens_out, n_token_capacity, n_token_count_out
)
_lib.llama_load_session_file.argtypes = [
llama_context_p,
c_char_p,
llama_token_p,
c_size_t,
c_size_t_p,
]
_lib.llama_load_session_file.restype = c_size_t
# LLAMA_API bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count);
def llama_save_session_file(
ctx: llama_context_p,
path_session: bytes,
tokens, # type: Array[llama_token]
n_token_count: c_size_t,
) -> int:
return _lib.llama_save_session_file(ctx, path_session, tokens, n_token_count)
_lib.llama_save_session_file.argtypes = [
llama_context_p,
c_char_p,
llama_token_p,
c_size_t,
]
_lib.llama_save_session_file.restype = c_size_t
# Run the llama inference to obtain the logits and probabilities for the next token.
# tokens + n_tokens is the provided batch of new tokens to process
# n_past is the number of tokens to use from previous eval calls
# Returns 0 on success
# LLAMA_API int llama_eval(
# struct llama_context * ctx,
# const llama_token * tokens,
# int n_tokens,
# int n_past,
# int n_threads);
def llama_eval(
ctx: llama_context_p,
tokens, # type: Array[llama_token]
n_tokens: c_int,
n_past: c_int,
n_threads: c_int,
) -> int:
return _lib.llama_eval(ctx, tokens, n_tokens, n_past, n_threads)
_lib.llama_eval.argtypes = [llama_context_p, llama_token_p, c_int, c_int, c_int]
_lib.llama_eval.restype = c_int
# // Same as llama_eval, but use float matrix input directly.
# LLAMA_API int llama_eval_embd(
# struct llama_context * ctx,
# const float * embd,
# int n_tokens,
# int n_past,
# int n_threads);
def llama_eval_embd(
ctx: llama_context_p,
embd, # type: Array[c_float]
n_tokens: c_int,
n_past: c_int,
n_threads: c_int,
) -> int:
return _lib.llama_eval_embd(ctx, embd, n_tokens, n_past, n_threads)
_lib.llama_eval_embd.argtypes = [llama_context_p, c_float_p, c_int, c_int, c_int]
_lib.llama_eval_embd.restype = c_int
# Convert the provided text into tokens.
# The tokens pointer must be large enough to hold the resulting tokens.
# Returns the number of tokens on success, no more than n_max_tokens
# Returns a negative number on failure - the number of tokens that would have been returned
# TODO: not sure if correct
# LLAMA_API int llama_tokenize(
# struct llama_context * ctx,
# const char * text,
# llama_token * tokens,
# int n_max_tokens,
# bool add_bos);
def llama_tokenize(
ctx: llama_context_p,
text: bytes,
tokens, # type: Array[llama_token]
n_max_tokens: c_int,
add_bos: c_bool,
) -> int:
return _lib.llama_tokenize(ctx, text, tokens, n_max_tokens, add_bos)
_lib.llama_tokenize.argtypes = [llama_context_p, c_char_p, llama_token_p, c_int, c_bool]
_lib.llama_tokenize.restype = c_int
# LLAMA_API int llama_tokenize_with_model(
# const struct llama_model * model,
# const char * text,
# llama_token * tokens,
# int n_max_tokens,
# bool add_bos);
def llama_tokenize_with_model(
model: llama_model_p,
text: bytes,
tokens, # type: Array[llama_token]
n_max_tokens: c_int,
add_bos: c_bool,
) -> int:
return _lib.llama_tokenize_with_model(model, text, tokens, n_max_tokens, add_bos)
# LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
def llama_n_vocab(ctx: llama_context_p) -> int:
return _lib.llama_n_vocab(ctx)
_lib.llama_n_vocab.argtypes = [llama_context_p]
_lib.llama_n_vocab.restype = c_int
# LLAMA_API int llama_n_ctx (const struct llama_context * ctx);
def llama_n_ctx(ctx: llama_context_p) -> int:
return _lib.llama_n_ctx(ctx)
_lib.llama_n_ctx.argtypes = [llama_context_p]
_lib.llama_n_ctx.restype = c_int
# LLAMA_API int llama_n_embd (const struct llama_context * ctx);
def llama_n_embd(ctx: llama_context_p) -> int:
return _lib.llama_n_embd(ctx)
_lib.llama_n_embd.argtypes = [llama_context_p]
_lib.llama_n_embd.restype = c_int
# LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
def llama_n_vocab_from_model(model: llama_model_p) -> int:
return _lib.llama_n_vocab_from_model(model)
_lib.llama_n_vocab_from_model.argtypes = [llama_model_p]
_lib.llama_n_vocab_from_model.restype = c_int
# LLAMA_API int llama_n_ctx_from_model (const struct llama_model * model);
def llama_n_ctx_from_model(model: llama_model_p) -> int:
return _lib.llama_n_ctx_from_model(model)
_lib.llama_n_ctx_from_model.argtypes = [llama_model_p]
_lib.llama_n_ctx_from_model.restype = c_int
# LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
def llama_n_embd_from_model(model: llama_model_p) -> int:
return _lib.llama_n_embd_from_model(model)
_lib.llama_n_embd_from_model.argtypes = [llama_model_p]
_lib.llama_n_embd_from_model.restype = c_int
# // Get the vocabulary as output parameters.
# // Returns number of results.
# LLAMA_API int llama_get_vocab(
# const struct llama_context * ctx,
# const char * * strings,
# float * scores,
# int capacity);
def llama_get_vocab(
ctx: llama_context_p,
strings, # type: Array[c_char_p] # type: ignore
scores, # type: Array[c_float] # type: ignore
capacity: c_int,
) -> int:
return _lib.llama_get_vocab(ctx, strings, scores, capacity)
_lib.llama_get_vocab.argtypes = [llama_context_p, c_char_p, c_float, c_int]
_lib.llama_get_vocab.restype = c_int
# LLAMA_API int llama_get_vocab_from_model(
# const struct llama_model * model,
# const char * * strings,
# float * scores,
# int capacity);
def llama_get_vocab_from_model(
model: llama_model_p,
strings, # type: Array[c_char_p] # type: ignore
scores, # type: Array[c_float] # type: ignore
capacity: c_int,
) -> int:
return _lib.llama_get_vocab_from_model(model, strings, scores, capacity)
# Token logits obtained from the last call to llama_eval()
# The logits for the last token are stored in the last row
# Can be mutated in order to change the probabilities of the next token
# Rows: n_tokens
# Cols: n_vocab
# LLAMA_API float * llama_get_logits(struct llama_context * ctx);
def llama_get_logits(
ctx: llama_context_p,
): # type: (...) -> Array[float] # type: ignore
return _lib.llama_get_logits(ctx)
_lib.llama_get_logits.argtypes = [llama_context_p]
_lib.llama_get_logits.restype = c_float_p
# Get the embeddings for the input
# shape: [n_embd] (1-dimensional)
# LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
def llama_get_embeddings(
ctx: llama_context_p,
): # type: (...) -> Array[float] # type: ignore
return _lib.llama_get_embeddings(ctx)
_lib.llama_get_embeddings.argtypes = [llama_context_p]
_lib.llama_get_embeddings.restype = c_float_p
# // Token Id -> String. Uses the vocabulary in the provided context
# LLAMA_API const char * llama_token_to_str(
# const struct llama_context * ctx,
# llama_token token);
def llama_token_to_str(ctx: llama_context_p, token: llama_token) -> bytes:
return _lib.llama_token_to_str(ctx, token)
_lib.llama_token_to_str.argtypes = [llama_context_p, llama_token]
_lib.llama_token_to_str.restype = c_char_p
# LLAMA_API const char * llama_token_to_str_with_model(
# const struct llama_model * model,
# llama_token token);
def llama_token_to_str_with_model(model: llama_model_p, token: llama_token) -> bytes:
return _lib.llama_token_to_str_with_model(model, token)
_lib.llama_token_to_str_with_model.argtypes = [llama_model_p, llama_token]
_lib.llama_token_to_str_with_model.restype = c_char_p
# Special tokens
# LLAMA_API llama_token llama_token_bos(); // beginning-of-sentence
def llama_token_bos() -> int:
return _lib.llama_token_bos()
_lib.llama_token_bos.argtypes = []
_lib.llama_token_bos.restype = llama_token
# LLAMA_API llama_token llama_token_eos(); // end-of-sentence
def llama_token_eos() -> int:
return _lib.llama_token_eos()
_lib.llama_token_eos.argtypes = []
_lib.llama_token_eos.restype = llama_token
# LLAMA_API llama_token llama_token_nl(); // next-line
def llama_token_nl() -> int:
return _lib.llama_token_nl()
_lib.llama_token_nl.argtypes = []
_lib.llama_token_nl.restype = llama_token
# Sampling functions
# @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
# LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
def llama_sample_repetition_penalty(
ctx: llama_context_p,
candidates, # type: _Pointer[llama_token_data_array]
last_tokens_data, # type: Array[llama_token]
last_tokens_size: c_int,
penalty: c_float,
):
return _lib.llama_sample_repetition_penalty(
ctx, candidates, last_tokens_data, last_tokens_size, penalty
)
_lib.llama_sample_repetition_penalty.argtypes = [
llama_context_p,
llama_token_data_array_p,
llama_token_p,
c_int,
c_float,
]
_lib.llama_sample_repetition_penalty.restype = None
# @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
# LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
def llama_sample_frequency_and_presence_penalties(
ctx: llama_context_p,
candidates, # type: _Pointer[llama_token_data_array]
last_tokens_data, # type: Array[llama_token]
last_tokens_size: c_int,
alpha_frequency: c_float,
alpha_presence: c_float,
):
return _lib.llama_sample_frequency_and_presence_penalties(
ctx,
candidates,
last_tokens_data,
last_tokens_size,
alpha_frequency,
alpha_presence,
)
_lib.llama_sample_frequency_and_presence_penalties.argtypes = [
llama_context_p,
llama_token_data_array_p,
llama_token_p,
c_int,
c_float,
c_float,
]
_lib.llama_sample_frequency_and_presence_penalties.restype = None
# /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
# /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
# /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
# /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
# /// @params smooth_factor Smooth factor between guidance logits and original logits. 1.0f means only use guidance logits. 0.0f means only original logits.
# LLAMA_API void llama_sample_classifier_free_guidance(
# struct llama_context * ctx,
# llama_token_data_array * candidates,
# struct llama_context * guidance_ctx,
# float scale,
# float smooth_factor);
def llama_sample_classifier_free_guidance(
ctx: llama_context_p,
candidates, # type: _Pointer[llama_token_data_array]
guidance_ctx: llama_context_p,
scale: c_float,
smooth_factor: c_float,
):
return _lib.llama_sample_classifier_free_guidance(
ctx, candidates, guidance_ctx, scale, smooth_factor
)
_lib.llama_sample_classifier_free_guidance.argtypes = [
llama_context_p,
llama_token_data_array_p,
llama_context_p,
c_float,
c_float,
]
_lib.llama_sample_classifier_free_guidance.restype = None
# @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
# LLAMA_API void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates);
def llama_sample_softmax(
ctx: llama_context_p, candidates # type: _Pointer[llama_token_data]
):
return _lib.llama_sample_softmax(ctx, candidates)
_lib.llama_sample_softmax.argtypes = [
llama_context_p,
llama_token_data_array_p,
]
_lib.llama_sample_softmax.restype = None
# @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
# LLAMA_API void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep);
def llama_sample_top_k(
ctx: llama_context_p,
candidates, # type: _Pointer[llama_token_data_array]
k: c_int,
min_keep: c_size_t,
):
return _lib.llama_sample_top_k(ctx, candidates, k, min_keep)
_lib.llama_sample_top_k.argtypes = [
llama_context_p,
llama_token_data_array_p,
c_int,
c_size_t,
]
_lib.llama_sample_top_k.restype = None
# @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
# LLAMA_API void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
def llama_sample_top_p(
ctx: llama_context_p,
candidates, # type: _Pointer[llama_token_data_array]
p: c_float,
min_keep: c_size_t,
):
return _lib.llama_sample_top_p(ctx, candidates, p, min_keep)
_lib.llama_sample_top_p.argtypes = [
llama_context_p,
llama_token_data_array_p,
c_float,
c_size_t,
]
_lib.llama_sample_top_p.restype = None
# @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
# LLAMA_API void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep);
def llama_sample_tail_free(