8000 Update llama.cpp · LOGp/llama-cpp-python@df9362e · GitHub
[go: up one dir, main page]

Skip to content

Commit df9362e

Browse files
committed
Update llama.cpp
1 parent 3af7b21 commit df9362e

File tree

4 files changed

+8
-8
lines changed

4 files changed

+8
-8
lines changed

llama_cpp/llama.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -233,7 +233,7 @@ def __init__(
233233
rope_scaling_type: Optional[int] = llama_cpp.LLAMA_ROPE_SCALING_UNSPECIFIED,
234234
rope_freq_base: float = 0.0,
235235
rope_freq_scale: float = 0.0,
236-
yarn_ext_factor: float = float("nan"),
236+
yarn_ext_factor: float = -1.0,
237237
yarn_attn_factor: float = 1.0,
238238
yarn_beta_fast: float = 32.0,
239239
yarn_beta_slow: float = 1.0,

llama_cpp/llama_cpp.py

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -315,11 +315,11 @@ class llama_model_params(Structure):
315315

316316

317317
# struct llama_context_params {
318-
# uint32_t seed; // RNG seed, -1 for random
319-
# uint32_t n_ctx; // text context, 0 = from model
320-
# uint32_t n_batch; // prompt processing maximum batch size
321-
# uint32_t n_threads; // number of threads to use for generation
322-
# uint32_t n_threads_batch; // number of threads to use for batch processing
318+
# uint32_t seed; // RNG seed, -1 for random
319+
# uint32_t n_ctx; // text context, 0 = from model
320+
# uint32_t n_batch; // prompt processing maximum batch size
321+
# uint32_t n_threads; // number of threads to use for generation
322+
# uint32_t n_threads_batch; // number of threads to use for batch processing
323323
# int8_t rope_scaling_type; // RoPE scaling type, from `enum llama_rope_scaling_type`
324324

325325
# // ref: https://github.com/ggerganov/llama.cpp/pull/2054

llama_cpp/server/app.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -93,7 +93,7 @@ class Settings(BaseSettings):
9393
default=0.0, description="RoPE frequency scaling factor"
9494
)
9595
yarn_ext_factor: float = Field(
96-
default=float("nan")
96+
default=-1.0
9797
)
9898
yarn_attn_factor: float = Field(
9999
default=1.0

vendor/llama.cpp

0 commit comments

Comments
 (0)
0