informers 1.0.3 → 1.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f5340da0bce9d55a0339fac6b8806f09119df3e89567ecb37a77e1a5921b8fa2
-  data.tar.gz: 66a9d275cb2999ad14ba1cfd900bdcbf9fdc3d26ce29387acdd74452bf2050ef
+  metadata.gz: a61f01755798e81a975641d60e5bfe09484ced7ce6a3453020c9978dc35b1942
+  data.tar.gz: 811f9c1dc4499ae7de8ebf8e02c0c4e98a0c0bc0af6aaca51025e42ba8165540
 SHA512:
-  metadata.gz: a4a0c3da3d8a3555a6f2debca8f2939b6536ac76386cdd6c7264890b2d00842d537ecfca352021fa349ff9c4636ba49c189f652a66676746d9ec2a8d97eecc2a
-  data.tar.gz: a06aa115b5966fd1b8da7a80d8481d3e61778f31c3bb0da143f329e81ae3f73d4a1d1b2ee01672f4e90742a35d68a23dd5c871c3b68ffad0c16d8e5de480a60f
+  metadata.gz: 97b27363fab1e43895e368dbddc819fd4db23d42ce517359e5971347cd902b654f0c66700f07b36cd5f476bd3ea205a91e4f7e7ee0e7d8d455f0dce377bedb2b
+  data.tar.gz: dd1a7f795609423419ce213b00a5aca409f6b4a5bffb111250b4deffcbc6a8113fadf8d603c59fa78fa0f310904a0a3299e3bcdc48101f574171a024d13567e6

data/CHANGELOG.md

@@ -1,3 +1,12 @@
+## 1.1.1 (2024-10-14)
+
+- Added `audio-classification` pipeline
+- Fixed error with `sentence-transformers/all-MiniLM-L6-v2`
+
+## 1.1.0 (2024-09-17)
+
+- Added more pipelines
+
 ## 1.0.3 (2024-08-29)
 
 - Added `model_output` option

data/README.md

@@ -229,17 +229,17 @@ result = model.(query, docs)
 
 ### Other
 
-You can use the feature extraction pipeline directly.
-
-```ruby
-model = Informers.pipeline("feature-extraction", "Xenova/all-MiniLM-L6-v2", quantized: false)
-embeddings = model.(sentences, pooling: "mean", normalize: true)
-```
-
 The model must include a `.onnx` file ([example](https://huggingface.co/Xenova/all-MiniLM-L6-v2/tree/main/onnx)). If the file is not at `onnx/model.onnx` or `onnx/model_quantized.onnx`, use the `model_file_name` option to specify the location.
 
 ## Pipelines
 
+- [Text](#text)
+- [Vision](#vision)
+- [Audio](#audio)
+- [Multimodel](#multimodal)
+
+### Text
+
 Embedding
 
 ```ruby
@@ -275,6 +275,48 @@ qa = Informers.pipeline("question-answering")
 qa.("Who invented Ruby?", "Ruby is a programming language created by Matz")
 ```
 
+Zero-shot classification
+
+```ruby
+classifier = Informers.pipeline("zero-shot-classification")
+classifier.("text", ["label1", "label2", "label3"])
+```
+
+Text generation
+
+```ruby
+generator = Informers.pipeline("text-generation")
+generator.("I enjoy walking with my cute dog,")
+```
+
+Text-to-text generation
+
+```ruby
+text2text = Informers.pipeline("text2text-generation")
+text2text.("translate from English to French: I'm very happy")
+```
+
+Translation
+
+```ruby
+translator = Informers.pipeline("translation", "Xenova/nllb-200-distilled-600M")
+translator.("जीवन एक चॉकलेट बॉक्स की तरह है।", src_lang: "hin_Deva", tgt_lang: "fra_Latn")
+```
+
+Summarization
+
+```ruby
+summarizer = Informers.pipeline("summarization")
+summarizer.("Many paragraphs of text")
+```
+
+Fill mask
+
+```ruby
+unmasker = Informers.pipeline("fill-mask")
+unmasker.("Paris is the [MASK] of France.")
+```
+
 Feature extraction
 
 ```ruby
@@ -282,6 +324,93 @@ extractor = Informers.pipeline("feature-extraction")
 extractor.("We are very happy to show you the 🤗 Transformers library.")
 ```
 
+### Vision
+
+Note: [ruby-vips](https://github.com/libvips/ruby-vips) is required to load images
+
+Image classification
+
+```ruby
+classifier = Informers.pipeline("image-classification")
+classifier.("image.jpg")
+```
+
+Zero-shot image classification
+
+```ruby
+classifier = Informers.pipeline("zero-shot-image-classification")
+classifier.("image.jpg", ["label1", "label2", "label3"])
+```
+
+Image segmentation
+
+```ruby
+segmenter = Informers.pipeline("image-segmentation")
+segmenter.("image.jpg")
+```
+
+Object detection
+
+```ruby
+detector = Informers.pipeline("object-detection")
+detector.("image.jpg")
+```
+
+Zero-shot object detection
+
+```ruby
+detector = Informers.pipeline("zero-shot-object-detection")
+detector.("image.jpg", ["label1", "label2", "label3"])
+```
+
+Depth estimation
+
+```ruby
+estimator = Informers.pipeline("depth-estimation")
+estimator.("image.jpg")
+```
+
+Image-to-image
+
+```ruby
+upscaler = Informers.pipeline("image-to-image")
+upscaler.("image.jpg")
+```
+
+Image feature extraction
+
+```ruby
+extractor = Informers.pipeline("image-feature-extraction")
+extractor.("image.jpg")
+```
+
+### Audio
+
+Note: [ffmpeg](https://www.ffmpeg.org/) is required to load audio files
+
+Audio classification
+
+```ruby
+classifier = Informers.pipeline("audio-classification")
+classifier.("audio.wav")
+```
+
+### Multimodal
+
+Image captioning
+
+```ruby
+captioner = Informers.pipeline("image-to-text")
+captioner.("image.jpg")
+```
+
+Document question answering
+
+```ruby
+qa = Informers.pipeline("document-question-answering")
+qa.("image.jpg", "What is the invoice number?")
+```
+
 ## Credits
 
 This library was ported from [Transformers.js](https://github.com/xenova/transformers.js) and is available under the same license.
@@ -321,5 +450,6 @@ To get started with development:
 git clone https://github.com/ankane/informers.git
 cd informers
 bundle install
+bundle exec rake download:files
 bundle exec rake test
 ```

data/lib/informers/configs.rb

@@ -1,17 +1,19 @@
 module Informers
   class PretrainedConfig
-    attr_reader :model_type, :problem_type, :id2label
-
     def initialize(config_json)
-      @is_encoder_decoder = false
-
-      @model_type = config_json["model_type"]
-      @problem_type = config_json["problem_type"]
-      @id2label = config_json["id2label"]
+      @config_json = config_json.to_h
     end
 
     def [](key)
-      instance_variable_get("@#{key}")
+      @config_json[key.to_s]
+    end
+
+    def []=(key, value)
+      @config_json[key.to_s] = value
+    end
+
+    def to_h
+      @config_json.to_h
     end
 
     def self.from_pretrained(

data/lib/informers/model.rb

@@ -1,19 +1,12 @@
 module Informers
   class Model
     def initialize(model_id, quantized: false)
-      @model_id = model_id
       @model = Informers.pipeline("embedding", model_id, quantized: quantized)
+      @options = model_id == "mixedbread-ai/mxbai-embed-large-v1" ? {pooling: "cls", normalize: false} : {}
     end
 
     def embed(texts)
-      case @model_id
-      when "sentence-transformers/all-MiniLM-L6-v2", "Xenova/all-MiniLM-L6-v2", "Xenova/multi-qa-MiniLM-L6-cos-v1", "Supabase/gte-small"
-        @model.(texts)
-      when "mixedbread-ai/mxbai-embed-large-v1"
-        @model.(texts, pooling: "cls", normalize: false)
-      else
-        raise Error, "Use the embedding pipeline for this model: #{@model_id}"
-      end
+      @model.(texts, **@options)
     end
   end
 end