aoororachain 0.1.4 → 0.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8fd799856d56a712154c7c6adf5c20e73e34ce7740aa70490dc84c37f7dd7905
-  data.tar.gz: e4d4a828187141c420cecff71b38c17fdd9556b9f486a21c336ee83a1b6b20d4
+  metadata.gz: 5f3616ebd00d6c06568a9b1c5d31a99147eb5f0afe48749f92e8275ee31c8272
+  data.tar.gz: 27711347c6c7306a16dace3cc74d931b2585fbfe953ef61166e7a51b38893dca
 SHA512:
-  metadata.gz: a157218dc07395105e4f6bda26f2a97df95d882b514a800401fb7a86596d9fe8583575395f4d16797f91800913c9b4fd639826001d5d1360c104fa0896da48e7
-  data.tar.gz: 554a6f6c5bbac2d094a0d35fa296feb6d5bd7fc999d6579d3d7d44c8161f52acc282b098bdcc6d387a3e5bcd74d2ae29a7602ce04509aa49948308a618d4052e
+  metadata.gz: 29096f58726afdfdc68f78f6effc57fcc8e79e645d2fccd51c87ca560d4f9ffed90c66873e9d9e69e795ec882ff6b02f12d951e5d4fd4406a9b7a11e64e56615
+  data.tar.gz: 7a67196e23436c158aa259cc2cb5c6e8c6e8cad6a5548ecbdfdd35cd9ca019b0d9e09378272140d9f893479be51fd48dd0f09bc956fe87a57d39345ea591b606

data/CHANGELOG.md

@@ -1,5 +1,8 @@
 ## [Unreleased]
 
-## [0.1.0] - 2023-05-24
+## [0.1.5] - 2023-07-10
+- Implements QA Retrieval with additional context.
+
+## [0.1.0] - 2023-06-24
 
 - Initial release

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    aoororachain (0.1.4)
+    aoororachain (0.1.5)
       chroma-db (~> 0.6.0)
       llm_client (~> 0.1.2)
       pdf-reader (~> 2.11)

data/README.md

@@ -138,24 +138,24 @@ collection_name = "ruby-documentation"
 # You can define a custom Parser to clean data and maybe extract metadata.
 # Here is the code of RubyDocParser that does exactly that.
 class RubyDocParser
-def self.parse(text)
-name_match = text.match(/Name (\w+)/)
-constant_match = text.match(/Constant (\w+)/)
-
-object_match = text.match(/Object (\w+)/)
-method_match = text.match(/Method ([\w\[\]\+\=\-\*\%\/]+)/)
-
-metadata = {}
-metadata[:name] = name_match[1] if name_match
-metadata[:constant] = constant_match[1] if constant_match
-metadata[:object] = object_match[1] if object_match
-metadata[:method] = method_match[1] if method_match
-metadata[:lang] = :ruby
-metadata[:version] = "3.2"
-
-text.gsub!(/\s+/, " ").strip!
-[text, metadata]
-end
+  def self.parse(text)
+    name_match = text.match(/Name (\w+)/)
+    constant_match = text.match(/Constant (\w+)/)
+    
+    object_match = text.match(/Object (\w+)/)
+    method_match = text.match(/Method ([\w\[\]\+\=\-\*\%\/]+)/)
+    
+    metadata = {}
+    metadata[:name] = name_match[1] if name_match
+    metadata[:constant] = constant_match[1] if constant_match
+    metadata[:object] = object_match[1] if object_match
+    metadata[:method] = method_match[1] if method_match
+    metadata[:lang] = :ruby
+    metadata[:version] = "3.2"
+    
+    text.gsub!(/\s+/, " ").strip!
+    [text, metadata]
+  end
 end
 
 # A DirectoryLoader points to a path and sets the glob for the files you want to load. 
@@ -170,7 +170,7 @@ text_splitter = Aoororachain::RecursiveTextSplitter.new(size: 512, overlap: 0)
 
 texts = []
 files.each do |file|
-texts.concat(text_splitter.split_documents(file))
+  texts.concat(text_splitter.split_documents(file))
 end
 
 # The final step is to create and store the embeddings.
@@ -191,7 +191,7 @@ vector_database.from_documents(texts, index: collection_name)
 With embedding loaded in the database, you can use a tool like Chroma UI -**not yet released** - to query documents.
 ![chroma-ui](https://github.com/mariochavez/aoororachain/assets/59967/d65dea13-c6ef-452a-9774-8cf3b47c048f)
 
-But it is more useful to query with Aoororachain.
+Now you can query your embeddings with Aoororachain.
 
 ```ruby
 # Define a retriever for the Vector database.
@@ -236,9 +236,29 @@ chain = Aoororachain::Chains::RetrievalQA.new(llm, retriever)
 # Create a template for the LLM. Aoororachain does not include any templates because these are model specific. The following template is for the Vicuna model.
 template = "A conversation between a human and an AI assistant. The assistant responds to a question using the context. Context: ===%{context}===. Question: %{prompt}"
 
-response = chain.complete(prompt: "how can I use the Data class to define a new class?", prompt_template: template)
+response = chain.complete(prompt: "given the following array [[1,3], [2,4]], how can I get a flatten and sorted array?", prompt_template: template)
+```
+
+_response_ is a Hash with two keys: _response_ and _sources_.
+
+```ruby
+pp response
+{:response=>
+  "User: Assistant: Assistant: To flatten the nested arrays in an array and sort it, you can use Ruby's built-in `sort` method along with the `flatten` method. Here is an example of how to do this for the given array [[1, 3], [2, 4]]:\n" +
+  "```ruby\n" +
+  "array = [[1, 3], [2, 4]]\n" +
+  "sorted_and_flattened_array = array.sort { |a, b| a[0] <=> b[0] }.flat_map(&:to_a)\n" +
+  "# Output: [1, 2, 3, 4]\n" +
+  "```\n",
+ :sources=>
+  [{"source"=>"./ruby-docs/hash-flatten.txt", "object"=>"Hash", "method"=>"flatten", "lang"=>"ruby", "version"=>"3.2"},
+   {"source"=>"./ruby-docs/array-flatten.txt", "object"=>"Array", "method"=>"flatten", "lang"=>"ruby", "version"=>"3.2"},
+   {"source"=>"./ruby-docs/array-flatten.txt", "object"=>"Array", "method"=>"flatten", "lang"=>"ruby", "version"=>"3.2"},
+   {"source"=>"./ruby-docs/array-flatten2.txt", "object"=>"Array", "method"=>"flatten", "lang"=>"ruby", "version"=>"3.2"}]}
 ```
 
+Where _response_ is tge generated response from the LLM and _sources_ is the list of text chunks that were sent to the LLM as context.
+
 ## Contributing
 
 Bug reports and pull requests are welcome on GitHub at https://github.com/mariochavez/aoororachain. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the [code of conduct](https://github.com/mariochavez/aoororachain/blob/main/CODE_OF_CONDUCT.md).

data/lib/aoororachain/chains/retrieval_qa.rb

@@ -9,22 +9,25 @@ module Aoororachain
         @type = type
       end
 
-      def complete(prompt:, prompt_template:)
+      def complete(prompt:, prompt_template:, additional_context: "")
         context = @retriever.search(prompt)
 
-        stuff_prompt = prompt_template % {context: context.map(&:document).join(" ").tr("\n", " "), prompt:}
+        mapped_context = context.map(&:document)
+        mapped_context << additional_context if !additional_context.nil? || additional_context != ""
+
+        stuff_prompt = prompt_template % {context: mapped_context.join(" ").tr("\n", " "), prompt:}
 
         success, response = @llm.complete(prompt: stuff_prompt)
 
         if success
           completion = {
-            response: response,
-            sources: context.map(&:metadata)
-          }
+            "sources" => context.map(&:metadata)
+          }.merge(response)
         else
           completion = {
-            response: "Sorry we had a problem with the LLM",
-            sources: []
+            "response" => "Sorry we had a problem with the LLM",
+            "sources" => [],
+            "model" => ""
           }
           Aoororachain::Util.log_error("Failed to complete", message: response)
         end

data/lib/aoororachain/llms/llama_server.rb

@@ -14,7 +14,7 @@ module Aoororachain
       def complete(prompt:)
         result = LlmClient.completion(prompt)
 
-        [result.success?, result.success? ? result.success.body["response"].gsub(/Usuario:.*Asistente:/, "") : result.failure.body]
+        [result.success?, result.success? ? result.success.body : result.failure.body]
       end
     end
   end

data/lib/aoororachain/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Aoororachain
-  VERSION = "0.1.4"
+  VERSION = "0.1.5"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: aoororachain
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.5
 platform: ruby
 authors:
 - Mario Alberto Chávez
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-07-04 00:00:00.000000000 Z
+date: 2023-07-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: chroma-db