hugging-face 0.1.0 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 81267c0c030b98028e201d0f52471d12a2aea6f6fe519354534522528e331d02
-  data.tar.gz: 235e128fdd8c7f534aace15ff606f89e432cb7fd120a44f94d3fc55cd4fa491a
+  metadata.gz: 93e213102e56d8e86de856912a1963f36e4b0a0d60f21c8a1f8d99b23a0f47d7
+  data.tar.gz: 877bd5a93e54ccbcb82943c19e4ac889657686aa656b0bd3bb911732b1d5d6ff
 SHA512:
-  metadata.gz: 7d620ad3f025f8f0ccae02a266b8ee74c41c019d50e76fbf3e24688a71f213cacd6c342cea23f9854522480c2f5945b4973a5ea58fba5f860bb16c8f2244971e
-  data.tar.gz: 1aa4a1fce880f38ae5367f168cb53ce3a93cf58450ab5d8d0a7ef6f5abe316efb713a1dcc97e9e592a91ba03484c915dc73dfe7401189fd8b8ba1b8cebfafc75
+  metadata.gz: ee534a915fa831884e42519f178ae709a08c201d79976d97b52319dd5322fc979c3dfe711a3e2210b0d53c0132490a089ccd2ca7c73730919f586582be0dc884
+  data.tar.gz: c4df146833ea986a86d9b9c060f594535b7bdf558dd9d28396dd100869c4a899dc4ddad1bdd2eba1ff8a09610ee4cada09a271adda9f4d6e58d21c5181c925a6

data/Gemfile

@@ -8,3 +8,4 @@ gemspec
 gem "rake", "~> 13.0"
 
 gem "rspec", "~> 3.0"
+gem "webmock", "~> 3.0"

data/Gemfile.lock

@@ -1,18 +1,25 @@
 PATH
   remote: .
   specs:
-    hugging-face (0.1.0)
+    hugging-face (0.2.0)
       faraday (~> 2.7)
 
 GEM
   remote: https://rubygems.org/
   specs:
+    addressable (2.8.4)
+      public_suffix (>= 2.0.2, < 6.0)
+    crack (0.4.5)
+      rexml
     diff-lcs (1.5.0)
     faraday (2.7.4)
       faraday-net_http (>= 2.0, < 3.1)
       ruby2_keywords (>= 0.0.4)
     faraday-net_http (3.0.2)
+    hashdiff (1.0.1)
+    public_suffix (5.0.1)
     rake (13.0.6)
+    rexml (3.2.5)
     rspec (3.12.0)
       rspec-core (~> 3.12.0)
       rspec-expectations (~> 3.12.0)
@@ -27,14 +34,20 @@ GEM
       rspec-support (~> 3.12.0)
     rspec-support (3.12.0)
     ruby2_keywords (0.0.5)
+    webmock (3.18.1)
+      addressable (>= 2.8.0)
+      crack (>= 0.3.2)
+      hashdiff (>= 0.4.0, < 2.0.0)
 
 PLATFORMS
   arm64-darwin-21
+  x86_64-linux
 
 DEPENDENCIES
   hugging-face!
   rake (~> 13.0)
   rspec (~> 3.0)
+  webmock (~> 3.0)
 
 BUNDLED WITH
    2.4.0

data/README.md

@@ -18,7 +18,36 @@ $ gem install hugging-face
 
 ## Usage
 
-TODO: Write usage instructions here
+```ruby
+require "hugging_face"
+```
+
+Instantiate a HuggigFace Inference API client:
+
+```ruby
+client = HuggingFace::InferenceApi.new(api_key: ENV['HUGGING_FACE_API_KEY'])
+```
+
+Question answering:
+
+```ruby
+client.question_answering(
+  question: 'What is my name?',
+  context: 'I am the only child. My father named his son John.'
+)
+```
+
+Text generation:
+
+```ruby
+client.text_generation(input: 'Can you please let us know more details about your ')
+```
+
+Summarization:
+
+```ruby
+client.summarization(input: 'The tower is 324 metres (1,063 ft) tall, about the same height as an 81-storey building, and the tallest structure in Paris. Its base is square, measuring 125 metres (410 ft) on each side. During its construction, the Eiffel Tower surpassed the Washington Monument to become the tallest man-made structure in the world, a title it held for 41 years until the Chrysler Building in New York City was finished in 1930.')
+```
 
 ## Development
 
@@ -33,3 +62,4 @@ Bug reports and pull requests are welcome on GitHub at https://github.com/alchap
 ## Code of Conduct
 
 Everyone interacting in the HuggingFace project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the [code of conduct](https://github.com/alchaplinsky/hugging-face/blob/main/CODE_OF_CONDUCT.md).
+

data/lib/hugging_face/base_api.rb

@@ -0,0 +1,46 @@
+require 'logger'
+require 'faraday'
+
+module HuggingFace
+  class BaseApi
+    HTTP_SERVICE_UNAVAILABLE = 503
+    JSON_CONTENT_TYPE = 'application/json'
+
+    def initialize(api_token:)
+      @headers = {
+        'Authorization' => 'Bearer ' + api_token,
+        'Content-Type' => JSON_CONTENT_TYPE
+      }
+    end
+
+    private
+
+    def build_connection(url)
+      Faraday.new(url, headers: @headers)
+    end
+
+    def request(connection:, input:)
+      response = connection.post { |req| req.body = input.to_json }
+
+      if response.success?
+        return parse_response response
+      else
+        raise ServiceUnavailable.new response.body if response.status == HTTP_SERVICE_UNAVAILABLE
+        raise Error.new response.body
+      end
+    end
+
+    def parse_response(response)
+      if response.headers['Content-Type'] == JSON_CONTENT_TYPE
+        JSON.parse(response.body)
+      else
+        response.body
+      end
+    end
+
+    def logger
+      @logger ||= Logger.new(STDOUT)
+    end
+  end
+end
+

data/lib/hugging_face/inference_api.rb

@@ -0,0 +1,64 @@
+module HuggingFace
+  class InferenceApi < BaseApi
+    HOST = "https://api-inference.huggingface.co"
+
+    # Retry connecting to the model for 1 minute
+    MAX_RETRY = 60
+
+    # Deafult models that can be overriden by 'model' param
+    QUESTION_ANSWERING_MODEL = 'distilbert-base-cased-distilled-squad'
+    SUMMARIZATION_MODEL = "sshleifer/distilbart-xsum-12-6"
+    GENERATION_MODEL = "distilgpt2"
+    EMBEDING_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
+
+    def call(input:, model:)
+      request(connection: connection(model), input: input)
+    end
+
+    def question_answering(question:, context:, model: QUESTION_ANSWERING_MODEL)
+      input = { question: question, context: context }
+
+      request connection: connection(model), input: input
+    end
+
+    def summarization(input:, model: SUMMARIZATION_MODEL)
+      request connection: connection(model), input: { inputs: input }
+    end
+
+    def text_generation(input:, model: GENERATION_MODEL)
+      request connection: connection(model), input: { inputs: input }
+    end
+
+    def embedding(input:)
+      request connection: connection(EMBEDING_MODEL), input: { inputs: input }
+    end
+
+    private
+
+    def connection(model)
+      if model == EMBEDING_MODEL
+        build_connection "#{HOST}/pipeline/feature-extraction/#{model}"
+      else
+        build_connection "#{HOST}/models/#{model}"
+      end
+    end
+
+    def request(connection:, input:)
+      retries = 0
+
+      begin
+        return super(connection: connection, input: input)
+      rescue ServiceUnavailable => exception
+
+        if retries < MAX_RETRY
+          logger.debug('Service unavailable, retrying...')
+          retries += 1
+          sleep 1
+          retry
+        else
+          raise exception
+        end
+      end
+    end
+  end
+end

data/lib/hugging_face/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HuggingFace
-  VERSION = "0.1.0"
+  VERSION = "0.3.0"
 end

data/lib/hugging_face.rb

@@ -1,9 +1,11 @@
 # frozen_string_literal: true
 
 require_relative "hugging_face/version"
-require_relative "hugging_face/interface_api"
+require_relative "hugging_face/base_api"
+require_relative "hugging_face/inference_api"
 
 module HuggingFace
   class Error < StandardError; end
+  class ServiceUnavailable < Error; end
   # Your code goes here...
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: hugging-face
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.3.0
 platform: ruby
 authors:
 - Alex Chaplinsky
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-05-15 00:00:00.000000000 Z
+date: 2023-05-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday
@@ -39,7 +39,8 @@ files:
 - README.md
 - Rakefile
 - lib/hugging_face.rb
-- lib/hugging_face/interface_api.rb
+- lib/hugging_face/base_api.rb
+- lib/hugging_face/inference_api.rb
 - lib/hugging_face/version.rb
 - sig/hugging_face.rbs
 homepage: https://rubygems.org/gems/hugging-face

data/lib/hugging_face/interface_api.rb

@@ -1,63 +0,0 @@
-require 'faraday'
-
-module HuggingFace
-  class InterfaceApi
-    HOST = "https://api-inference.huggingface.co"
-    MAX_RETRY = 2
-    HTTP_SEVICE_UNAVAILABLE = 503
-
-    QUESTION_ANSWERING_MODEL = 'distilbert-base-cased-distilled-squad'
-    SUMMARIZATION_MODEL = "sshleifer/distilbart-xsum-12-6"
-    GENERATION_MODEL = "distilgpt2"
-
-    def initialize(api_token:)
-      @headers = {
-        'Authorization' => 'Bearer ' + api_token,
-        'Content-Type' => 'application/json'
-      }
-    end
-
-    def call(input:, model:)
-      request(connection: connection(model), input: input)
-    end
-
-    def question_answering(question:, context:, model: QUESTION_ANSWERING_MODEL)
-      input = { question: question, context: context }
-
-      request(connection: connection(model), input: input)
-    end
-
-    def summarization(input:, model: SUMMARIZATION_MODEL)
-      request(connection: connection(model), input: { inputs: input })
-    end
-
-    def text_generation(input:, model: GENERATION_MODEL)
-      request(connection: connection(model), input: { inputs: input })
-    end
-
-    private
-
-    def request(connection:, input:)
-      retries = 0
-      while retries < MAX_RETRY
-        response = connection.post { |req| req.body = input.to_json }
-
-        break if response.success?
-
-        if response.status == HTTP_SEVICE_UNAVAILABLE
-          retries += 1
-          sleep 1
-          redo
-        end
-
-        raise "Error: #{response.body}"
-      end
-
-      return JSON.parse(response.body)
-    end
-
-    def connection(model)
-      Faraday.new(url: "#{HOST}/models/#{model}" , headers: @headers)
-    end
-  end
-end