Update Submodule vendor/llama.cpp 81df3f7..82209ef

Author: JamePeng

llama_cpp/_internals.py

@@ -758,10 +758,6 @@ def reset_timings(self):
     def print_timings(self):
         llama_cpp.llama_perf_context_print(self.ctx)
 
-    def print_memory_breakdown(self):
-        """print a breakdown of per-device memory use via LLAMA_LOG"""
-        llama_cpp.llama_memory_breakdown_print(self.ctx)
-
     # LoRA / ALoRA Dynamic Routing Methods
 
     def clear_loras(self):

llama_cpp/llama_cpp.py

@@ -1519,53 +1519,6 @@ class llama_params_fit_status(enum.IntEnum):
     LLAMA_PARAMS_FIT_STATUS_ERROR   = 2
 
 
-# // fits mparams and cparams to free device memory (assumes system memory is unlimited)
-# //   - returns true if the parameters could be successfully modified to fit device memory
-# //   - this function is NOT thread safe because it modifies the global llama logger state
-# //   - only parameters that have the same value as in llama_default_model_params are modified
-# //     with the exception of the context size which is modified if and only if equal to 0
-# LLAMA_API enum llama_params_fit_status llama_params_fit(
-#                                 const char   * path_model,
-#                 struct llama_model_params   * mparams,
-#                 struct llama_context_params * cparams,
-#                                         float * tensor_split,          // writable buffer for tensor split, needs at least llama_max_devices elements
-#     struct llama_model_tensor_buft_override * tensor_buft_overrides, // writable buffer for overrides, needs at least llama_max_tensor_buft_overrides elements
-#                                         size_t   margin,                // margin of memory to leave per device in bytes
-#                                     uint32_t   n_ctx_min,             // minimum context size to set when trying to reduce memory use
-#                         enum ggml_log_level   log_level);            // minimum log level to print during fitting, lower levels go to debug log
-@ctypes_function(
-    "llama_params_fit",
-    [
-        ctypes.c_char_p,
-        llama_model_params_p,
-        llama_context_params_p,
-        ctypes.POINTER(ctypes.c_float),
-        ctypes.POINTER(llama_model_tensor_buft_override),
-        ctypes.c_size_t,
-        ctypes.c_uint32,
-        ctypes.c_int,
-    ],
-    ctypes.c_int,
-)
-def llama_params_fit(
-    path_model: ctypes.c_char_p,
-    mparams: CtypesPointer[llama_model_params],
-    cparams: CtypesPointer[llama_context_params],
-    tensor_split: CtypesPointer[ctypes.c_float],
-    tensor_buft_overrides: CtypesPointer[llama_model_tensor_buft_override],
-    margin: ctypes.c_size_t,
-    n_ctx_min: ctypes.c_uint32,
-    log_level: int,
-    /,
-) -> int:
-    """
-    fits mparams and cparams to free device memory (assumes system memory is unlimited)
-    returns true if the parameters could be successfully modified to fit device memory
-    this function is NOT thread safe because it modifies the global llama logger state
-    """
-    ...
-
-
 # LLAMA_API int64_t llama_time_us(void);
 @ctypes_function(
     "llama_time_us",
@@ -4930,17 +4883,6 @@ def llama_perf_sampler_reset(chain: llama_sampler_p, /):
     ...
 
 
-# // print a breakdown of per-device memory use via LLAMA_LOG:
-# LLAMA_API void llama_memory_breakdown_print(const struct llama_context * ctx);
-@ctypes_function(
-    "llama_memory_breakdown_print",
-    [llama_context_p_ctypes],
-    None,
-)
-def llama_memory_breakdown_print(ctx: llama_context_p, /):
-    ...
-
-
 # //
 # // training
 # //

llama_cpp/mtmd_cpp.py

@@ -94,10 +94,19 @@ class mtmd_input_chunk_type(enum.IntEnum):
 mtmd_bitmap_p = NewType("mtmd_bitmap_p", int)
 mtmd_bitmap_p_ctypes = c_void_p
 
+# // position indexing for decoder model
+# enum mtmd_pos_type {
+#     MTMD_POS_TYPE_NORMAL, // number of positions equals to number of tokens
+#     MTMD_POS_TYPE_MROPE, // qwen-vl mrope style, each image takes max(t,h,w) position indexes
+# };
+class mtmd_pos_type(enum.IntEnum):
+    MTMD_POS_TYPE_NORMAL = 0  # number of positions equals to number of tokens
+    MTMD_POS_TYPE_MROPE  = 1  # qwen-vl mrope style, each image takes max(t,h,w) position indexes
+
 # struct mtmd_image_tokens {
 #     uint32_t nx; // number of tokens in x direction
 #     uint32_t ny; // number of tokens in y direction
-#     bool use_mrope_pos = false; // use M-RoPE position counting (the whole image is 1 temporal position)
+#     mtmd_pos_type pos = MTMD_POS_TYPE_NORMAL;
 #     uint32_t n_tokens() const { return nx * ny; }
 #     clip_image_f32_batch batch_f32; // preprocessed image patches
 #     std::string id; // optional user-defined ID, useful for KV cache tracking
@@ -269,36 +278,37 @@ def mtmd_free(ctx: mtmd_context_p):
     ...
 
 # // whether we need to set non-causal mask before llama_decode
-# MTMD_API bool mtmd_decode_use_non_causal(mtmd_context * ctx);
+# // if chunk is nullptr, we assume the default case where chunk is an image chunk
+# MTMD_API bool mtmd_decode_use_non_causal(const mtmd_context * ctx, const mtmd_input_chunk * chunk);
 @ctypes_function_mtmd(
-    "mtmd_decode_use_non_causal", [mtmd_context_p_ctypes], c_bool)
-def mtmd_decode_use_non_causal(ctx: mtmd_context_p) -> c_bool:
+    "mtmd_decode_use_non_causal", [mtmd_context_p_ctypes, mtmd_input_chunk_p_ctypes], c_bool)
+def mtmd_decode_use_non_causal(ctx: mtmd_context_p, chunk: mtmd_input_chunk_p) -> c_bool:
     ...
 
 # // whether the current model use M-RoPE for llama_decode
-# MTMD_API bool mtmd_decode_use_mrope(mtmd_context * ctx);
+# MTMD_API bool mtmd_decode_use_mrope(const mtmd_context * ctx);
 @ctypes_function_mtmd(
     "mtmd_decode_use_mrope", [mtmd_context_p_ctypes], c_bool)
 def mtmd_decode_use_mrope(ctx: mtmd_context_p) -> c_bool:
     ...
 
 # // whether the current model supports vision input
-# MTMD_API bool mtmd_support_vision(mtmd_context * ctx);
+# MTMD_API bool mtmd_support_vision(const mtmd_context * ctx);
 @ctypes_function_mtmd(
     "mtmd_support_vision", [mtmd_context_p_ctypes], c_bool)
 def mtmd_support_vision(ctx: mtmd_context_p) -> c_bool:
     ...
 
 # // whether the current model supports audio input
-# MTMD_API bool mtmd_support_audio(mtmd_context * ctx);
+# MTMD_API bool mtmd_support_audio(const mtmd_context * ctx);
 @ctypes_function_mtmd(
     "mtmd_support_audio", [mtmd_context_p_ctypes], c_bool)
 def mtmd_support_audio(ctx: mtmd_context_p) -> c_bool:
     ...
 
 # // get audio sample rate in Hz, for example 16000 for Whisper
 # // return -1 if audio is not supported
-# MTMD_API int mtmd_get_audio_sample_rate(mtmd_context * ctx);
+# MTMD_API int mtmd_get_audio_sample_rate(const mtmd_context * ctx);
 @ctypes_function_mtmd(
     "mtmd_get_audio_sample_rate", [mtmd_context_p_ctypes], c_int)
 def mtmd_get_audio_sample_rate(ctx: mtmd_context_p) -> c_int: