8000 feat: Update llama.cpp · coderonion/llama-cpp-python@7ba257e · GitHub
[go: up one dir, main page]

Skip to content

Commit 7ba257e

Browse files
committed
feat: Update llama.cpp
1 parent f3fb90b commit 7ba257e

File tree

2 files changed

+33
-1
lines changed

2 files changed

+33
-1
lines changed

llama_cpp/llama_cpp.py

Lines changed: 32 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -221,6 +221,7 @@
221221
# LLAMA_VOCAB_PRE_TYPE_GPT3_FINNISH = 24,
222222
# LLAMA_VOCAB_PRE_TYPE_EXAONE = 25,
223223
# LLAMA_VOCAB_PRE_TYPE_CHAMELEON = 26,
224+
# LLAMA_VOCAB_PRE_TYPE_MINERVA = 27,
224225
# };
225226
LLAMA_VOCAB_PRE_TYPE_DEFAULT = 0
226227
LLAMA_VOCAB_PRE_TYPE_LLAMA3 = 1
@@ -249,6 +250,7 @@
249250
LLAMA_VOCAB_PRE_TYPE_GPT3_FINNISH = 24
250251
LLAMA_VOCAB_PRE_TYPE_EXAONE = 25
251252
LLAMA_VOCAB_PRE_TYPE_CHAMELEON = 26
253+
LLAMA_VOCAB_PRE_TYPE_MINERVA = 27
252254

253255

254256
# // note: these values should be synchronized with ggml_rope
@@ -392,12 +394,14 @@
392394
# LLAMA_ROPE_SCALING_TYPE_NONE = 0,
393395
# LLAMA_ROPE_SCALING_TYPE_LINEAR = 1,
394396
# LLAMA_ROPE_SCALING_TYPE_YARN = 2,
397+
# LLAMA_ROPE_SCALING_TYPE_LONGROPE = 3,
395398
# LLAMA_ROPE_SCALING_TYPE_MAX_VALUE = LLAMA_ROPE_SCALING_TYPE_YARN,
396399
# };
397400
LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED = -1
398401
LLAMA_ROPE_SCALING_TYPE_NONE = 0
399402
LLAMA_ROPE_SCALING_TYPE_LINEAR = 1
400403
LLAMA_ROPE_SCALING_TYPE_YARN = 2
404+
LLAMA_ROPE_SCALING_TYPE_LONGROPE = 3
401405
LLAMA_ROPE_SCALING_TYPE_MAX_VALUE = LLAMA_ROPE_SCALING_TYPE_YARN
402406

403407
# enum llama_pooling_type {
@@ -2933,6 +2937,34 @@ def llama_chat_apply_template(
29332937
...
29342938

29352939

2940+
# // Get list of built-in chat templates
2941+
# LLAMA_API int32_t llama_chat_builtin_templates(const char ** output, size_t len);
2942+
@ctypes_function(
2943+
"llama_chat_builtin_templates",
2944+
[
2945+
ctypes.POINTER(ctypes.c_char_p),
2946+
ctypes.c_size_t,
2947+
],
2948+
ctypes.c_int32,
2949+
)
2950+
def llama_chat_builtin_templates(
2951+
output: CtypesArray[bytes],
2952+
len: Union[ctypes.c_size_t, int],
2953+
/,
2954+
) -> int:
2955+
"""Get list of built-in chat templates.
2956+
2957+
Args:
2958+
output: Output buffer to store template names.
2959+
len: Length of the output buffer.
2960+
2961+
Returns:
2962+
Number of templates available.
2963+
Returns a negative number on error.
2964+
"""
2965+
...
2966+
2967+
29362968
# //
29372969
# // Sampling API
29382970
# //

vendor/llama.cpp

0 commit comments

Comments
 (0)
0