llama_cpp 0.17.8 → 0.17.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d4283cb48aef0a09e98f7e065a4e1f7fdd9532a334c7e184149d4b9ff4593927
-  data.tar.gz: 92db5cafd75f507c8748548ecf12697356ba4b8de08a0f6f9fdb9ee790757883
+  metadata.gz: ec7b41006345cfe57e355d207718ae460a5a2c519d67027daddb7d40ab346754
+  data.tar.gz: efcfa7eef5374790cef788f719bede1ab3cd30e5dec79be0ba9844bfbcd433c2
 SHA512:
-  metadata.gz: 9d18d4bbfe834cfb4c16a86bc231421fc038a5560cb7a2f5fb436db89a418029373057da6a2ad2331830b970df0a90453d4a8da9181ce39dadcb6957172f0d1a
-  data.tar.gz: 3976c8a29227e9cd37e91dee00b9a9ad7fcad3716ba3cda9c6f15773de2ec8ee112d00149717ac89c10c0a198f06e046d9be6cc2cbb1ef45e8fac5403ed4dcab
+  metadata.gz: c39150e77652060fcfee2c73b1334ff53ed40caf58679efa727a0d0d54bffe036b8795be9751cef8517b5d4b2300fb77fa5753a9c94b803988ec429b999591c2
+  data.tar.gz: 6054dee2f39ac2690de2836405ac793bf987538ef0780cffab2700544bc60ceaca8a4b109407c66dcf2dd9f38ffe6b98659d7ca4ddb5c9e93fa85d9884cea26e

data/CHANGELOG.md

@@ -1,3 +1,17 @@
+## [[0.17.10](https://github.com/yoshoku/llama_cpp.rb/compare/v0.17.9...v0.17.10)] - 2024-09-07
+
+- Change supported llama.cpp version to b3676.
+  - Add `LLAMA_VOCAB_TYPE_RWKV` constant.
+  - Add `LLAMA_FTYPE_MOSTLY_TQ1_0` and `LLAMA_FTYPE_MOSTLY_TQ2_0` constants.
+  - Change type of n_threads and n_threads_batch from uint32_t to int32 in native extension codes.
+
+Implementation bindings for llama_attach_threadpool and llama_detach_threadpool have been skipped.
+
+## [[0.17.9](https://github.com/yoshoku/llama_cpp.rb/compare/v0.17.8...v0.17.9)] - 2024-08-31
+
+- Change supported llama.cpp version to b3639.
+  - There are no changes in the API.
+
 ## [[0.17.8](https://github.com/yoshoku/llama_cpp.rb/compare/v0.17.7...v0.17.8)] - 2024-08-25
 
 - Change supported llama.cpp version to b3614.

data/ext/llama_cpp/llama_cpp.cpp

@@ -2512,7 +2512,7 @@ private:
       rb_raise(rb_eArgError, "LLaMA context is not initialized");
       return Qnil;
     }
-    llama_set_n_threads(ptr->ctx, NUM2UINT(n_threads), NUM2UINT(n_threads_batch));
+    llama_set_n_threads(ptr->ctx, NUM2INT(n_threads), NUM2INT(n_threads_batch));
     return Qnil;
   }
 
@@ -2558,7 +2558,7 @@ private:
       rb_raise(rb_eRuntimeError, "LLaMA context is not initialized");
       return Qnil;
     }
-    return UINT2NUM(llama_n_threads(ptr->ctx));
+    return INT2NUM(llama_n_threads(ptr->ctx));
   }
 
   static VALUE _llama_context_n_threads_batch(VALUE self) {
@@ -2567,7 +2567,7 @@ private:
       rb_raise(rb_eRuntimeError, "LLaMA context is not initialized");
       return Qnil;
     }
-    return UINT2NUM(llama_n_threads_batch(ptr->ctx));
+    return INT2NUM(llama_n_threads_batch(ptr->ctx));
   }
 
   static VALUE _llama_context_get_timings(VALUE self) {
@@ -3611,6 +3611,7 @@ extern "C" void Init_llama_cpp(void) {
   rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_TYPE_BPE", INT2NUM(LLAMA_VOCAB_TYPE_BPE));
   rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_TYPE_WPM", INT2NUM(LLAMA_VOCAB_TYPE_WPM));
   rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_TYPE_UGM", INT2NUM(LLAMA_VOCAB_TYPE_UGM));
+  rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_TYPE_RWKV", INT2NUM(LLAMA_VOCAB_TYPE_RWKV));
 
   rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_DEFAULT", INT2NUM(LLAMA_VOCAB_PRE_TYPE_DEFAULT));
   rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_LLAMA3", INT2NUM(LLAMA_VOCAB_PRE_TYPE_LLAMA3));
@@ -3690,6 +3691,8 @@ extern "C" void Init_llama_cpp(void) {
   rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_Q4_0_4_4", INT2NUM(LLAMA_FTYPE_MOSTLY_Q4_0_4_4));
   rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_Q4_0_4_8", INT2NUM(LLAMA_FTYPE_MOSTLY_Q4_0_4_8));
   rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_Q4_0_8_8", INT2NUM(LLAMA_FTYPE_MOSTLY_Q4_0_8_8));
+  rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_TQ1_0", INT2NUM(LLAMA_FTYPE_MOSTLY_TQ1_0));
+  rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_TQ2_0", INT2NUM(LLAMA_FTYPE_MOSTLY_TQ2_0));
 
   rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_GUESSED", INT2NUM(LLAMA_FTYPE_GUESSED));
 

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 = '0.17.8'
+  VERSION = '0.17.10'
 
   # The supported version of llama.cpp.
-  LLAMA_CPP_VERSION = 'b3614'
+  LLAMA_CPP_VERSION = 'b3676'
 end

data/sig/llama_cpp.rbs

@@ -16,6 +16,7 @@ module LLaMACpp
   LLAMA_VOCAB_TYPE_BPE: Integer
   LLAMA_VOCAB_TYPE_WPM: Integer
   LLAMA_VOCAB_TYPE_UGM: Integer
+  LLAMA_VOCAB_TYPE_RWKV: Integer
 
   LLAMA_VOCAB_PRE_TYPE_DEFAULT: Integer
   LLAMA_VOCAB_PRE_TYPE_LLAMA3: Integer
@@ -87,6 +88,8 @@ module LLaMACpp
   LLAMA_FTYPE_MOSTLY_Q4_0_4_4: Integer
   LLAMA_FTYPE_MOSTLY_Q4_0_4_8: Integer
   LLAMA_FTYPE_MOSTLY_Q4_0_8_8: Integer
+  LLAMA_FTYPE_MOSTLY_TQ1_0: Integer
+  LLAMA_FTYPE_MOSTLY_TQ2_0: Integer
   LLAMA_FTYPE_GUESSED: Integer
 
   LLAMA_KV_OVERRIDE_TYPE_INT: Integer

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: llama_cpp
 version: !ruby/object:Gem::Version
-  version: 0.17.8
+  version: 0.17.10
 platform: ruby
 authors:
 - yoshoku
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-08-25 00:00:00.000000000 Z
+date: 2024-09-07 00:00:00.000000000 Z
 dependencies: []
 description: llama_cpp.rb provides Ruby bindings for the llama.cpp.
 email: