llama_cpp 0.0.6 → 0.0.7

data/sig/llama_cpp.rbs

@@ -12,9 +12,12 @@ module LLaMACpp
   LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16: Integer
   LLAMA_FTYPE_MOSTLY_Q4_2: Integer
   LLAMA_FTYPE_MOSTLY_Q4_3: Integer
+  LLAMA_FTYPE_MOSTLY_Q8_0: Integer
+  LLAMA_FTYPE_MOSTLY_Q5_0: Integer
+  LLAMA_FTYPE_MOSTLY_Q5_1: Integer
 
   def self?.model_quantize: (input_path: String, output_path: String, ftype: Integer, ?n_threads: Integer) -> void
-  def self?.generate: (::LLaMACpp::Context, String, ?n_threads: Integer) -> String
+  def self?.generate: (::LLaMACpp::Context, String, ?n_predict: Integer, ?n_threads: Integer) -> String
   def self?.print_system_info: () -> void
   def self?.token_bos: () -> Integer
   def self?.token_eos: () -> Integer
@@ -27,6 +30,7 @@ module LLaMACpp
     def initialize: (model_path: String, params: ::LLaMACpp::ContextParams) -> void
                   | () -> void
     def embeddings: () -> Array[Float]
+    def empty?: () -> bool
     def eval: (tokens: Array[Integer], n_past: Integer, ?n_tokens: Integer, ?n_threads: Integer) -> void
     def free: () -> void
     def load: (model_path: String, params: ::LLaMACpp::ContextParams) -> void
@@ -59,9 +63,20 @@ module LLaMACpp
     def seed=: (Integer) -> Integer
     def use_mlock: () -> bool
     def use_mlock=: (bool) -> bool
+    def use_mmap: () -> bool
+    def use_mmap=: (bool) -> bool
     def vocab_only: () -> bool
     def vocab_only=: (bool) -> bool
   end
 
   class Params = ContextParams
+
+  class Client
+    def initialize(model_path: String, ?lora_adapter_path: String, ?lora_base_path: String,
+                   ?n_ctx: Integer, ?n_parts: Integer, ?memory_f16: bool, ?use_mmap: bool, ?use_mlock: bool,
+                   ?embedding: bool, ?n_threads: Integer, ?seed: Integer) -> void
+    def completions(String, ?max_tokens: Integer, ?n_keep: Integer, ?repeat_last_n: Integer, ?n_batch: Integer,
+                    ?top_k: Integer, ?top_p: Float, ?temperature: Float, ?repeat_penalty: Float) -> String
+    def embeddings(String) -> Array[Float]
+  end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: llama_cpp
 version: !ruby/object:Gem::Version
-  version: 0.0.6
+  version: 0.0.7
 platform: ruby
 authors:
 - yoshoku
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-04-22 00:00:00.000000000 Z
+date: 2023-04-29 00:00:00.000000000 Z
 dependencies: []
 description: llama_cpp.rb provides Ruby bindings for the llama.cpp.
 email:
@@ -27,12 +27,15 @@ files:
 - ext/llama_cpp/llama_cpp.h
 - ext/llama_cpp/src/LICENSE
 - ext/llama_cpp/src/ggml-cuda.h
+- ext/llama_cpp/src/ggml-opencl.c
+- ext/llama_cpp/src/ggml-opencl.h
 - ext/llama_cpp/src/ggml.c
 - ext/llama_cpp/src/ggml.h
 - ext/llama_cpp/src/llama.cpp
 - ext/llama_cpp/src/llama.h
 - ext/llama_cpp/src/llama_util.h
 - lib/llama_cpp.rb
+- lib/llama_cpp/client.rb
 - lib/llama_cpp/version.rb
 - sig/llama_cpp.rbs
 homepage: https://github.com/yoshoku/llama_cpp.rb