llama_cpp 0.3.5 → 0.3.7

data/ext/llama_cpp/src/llama.h

@@ -86,7 +86,20 @@ extern "C" {
 
     typedef void (*llama_progress_callback)(float progress, void *ctx);
 
-   struct llama_context_params {
+    enum llama_log_level {
+        LLAMA_LOG_LEVEL_ERROR = 2,
+        LLAMA_LOG_LEVEL_WARN  = 3,
+        LLAMA_LOG_LEVEL_INFO  = 4
+    };
+
+    // Signature for logging events
+    // Note that text includes the new line character at the end for most events.
+    // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
+    // if it exists.
+    // It might not exist for progress report where '.' is output repeatedly.
+    typedef void (*llama_log_callback)(llama_log_level level, const char * text, void * user_data);
+
+    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
@@ -108,6 +121,7 @@ extern "C" {
 
         // 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 mul_mat_q;  // if true, use experimental mul_mat_q kernels
         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
@@ -194,6 +208,10 @@ extern "C" {
         int32_t n_eval;
     };
 
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
+
     LLAMA_API int llama_max_devices();
 
     LLAMA_API struct llama_context_params llama_context_default_params();

data/lib/llama_cpp/version.rb

@@ -3,8 +3,8 @@
 # llama_cpp.rb provides Ruby bindings for the llama.cpp.
 module LLaMACpp
   # The version of llama_cpp.rb you install.
-  VERSION = '0.3.5'
+  VERSION = '0.3.7'
 
   # The version of llama.cpp bundled with llama_cpp.rb.
-  LLAMA_CPP_VERSION = 'master-1a94186'
+  LLAMA_CPP_VERSION = 'master-9ca4abe'
 end

data/sig/llama_cpp.rbs

@@ -163,6 +163,8 @@ module LLaMACpp
     def rope_freq_scale: () -> Float
     def low_vram: () -> bool
     def low_vram=: (bool) -> bool
+    def mul_mat_q: () -> bool
+    def mul_mat_q=: (bool) -> bool
     def seed: () -> Integer
     def seed=: (Integer) -> Integer
     def use_mlock: () -> bool

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: llama_cpp
 version: !ruby/object:Gem::Version
-  version: 0.3.5
+  version: 0.3.7
 platform: ruby
 authors:
 - yoshoku
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-07-29 00:00:00.000000000 Z
+date: 2023-08-11 00:00:00.000000000 Z
 dependencies: []
 description: llama_cpp.rb provides Ruby bindings for the llama.cpp.
 email:
@@ -30,6 +30,8 @@ files:
 - ext/llama_cpp/llama_cpp.cpp
 - ext/llama_cpp/llama_cpp.h
 - ext/llama_cpp/src/LICENSE
+- ext/llama_cpp/src/ggml-alloc.c
+- ext/llama_cpp/src/ggml-alloc.h
 - ext/llama_cpp/src/ggml-cuda.cu
 - ext/llama_cpp/src/ggml-cuda.h
 - ext/llama_cpp/src/ggml-metal.h