langchainrb 0.11.2 → 0.11.3

Sign up to get free protection for your applications and to get access to all the features.
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: 0d2eb76b006864583607672a93cd81a5c37b259a34a2760f7d686f8923b60b22
4
- data.tar.gz: 4e307d737f9e519e68c6adb632ff863ca365907789183756fcabe7bf0a79df4f
3
+ metadata.gz: c678ae75bc25b0501223f5b6ffd396a9159af4d0ddd87ddb1657429ed2ba24ce
4
+ data.tar.gz: df50ef0a6d9c1a3100153a06084556cac983069d1a38739bd6606f39f63bd332
5
5
  SHA512:
6
- metadata.gz: 757e6fd5733b0365eb5b4309589f8dfab8659f451a8f31e922c7ded5ca1d974ab6d6a9b8178aa773f2b812058ea51db2f1560d18215beb4364c15b953f57648d
7
- data.tar.gz: 8c7ffe8b7f94c32a0e1afedff943871a097811266981e5cb0e3353b66f7eebdb387565f3d2292dda665d69dac2ad29584cad14c07c90b9c2def768ec93a403ea
6
+ metadata.gz: 3ec9f92f4c6221184b7a0a2c118caa6a56e7bc8505a83d5b5acb4daeb769ff90d8822b43c28f57adc06435dd2df5577268721345c4061d3dad6ecb919be18efc
7
+ data.tar.gz: 53d54b0c6a82082438f2e2f1ca70d097a9b916bc283b72e52ce466b6f012c9624cf094586b17e93eaa49a796bf9911051d3f4b494b9ecc93c3ac6ee6cdc7e8fe
data/CHANGELOG.md CHANGED
@@ -1,4 +1,12 @@
1
1
  ## [Unreleased]
2
+ - New `Langchain::Processors::Pptx` to parse .pptx files
3
+ - New `Langchain::LLM::Anthropic#chat()` support
4
+ - Misc fixes
5
+
6
+ ## [0.11.3]
7
+ - New `Langchain::Processors::Pptx` to parse .pptx files
8
+ - New `Langchain::LLM::Anthropic#chat()` support
9
+ - Misc fixes
2
10
 
3
11
  ## [0.11.2]
4
12
  - New `Langchain::Assistant#clear_thread!` and `Langchain::Assistant#instructions=` methods
data/README.md CHANGED
@@ -59,7 +59,7 @@ Langchain.rb wraps supported LLMs in a unified interface allowing you to easily
59
59
  | -------- |:------------------:| :-------: | :-----------------: | :-------: | :----------------- |
60
60
  | [OpenAI](https://openai.com/?utm_source=langchainrb&utm_medium=github) | ✅ | ✅ | ✅ | ❌ | Including Azure OpenAI |
61
61
  | [AI21](https://ai21.com/?utm_source=langchainrb&utm_medium=github) | ❌ | ✅ | ❌ | ✅ | |
62
- | [Anthropic](https://anthropic.com/?utm_source=langchainrb&utm_medium=github) | ❌ | ✅ | | ❌ | |
62
+ | [Anthropic](https://anthropic.com/?utm_source=langchainrb&utm_medium=github) | ❌ | ✅ | | ❌ | |
63
63
  | [AWS Bedrock](https://aws.amazon.com/bedrock?utm_source=langchainrb&utm_medium=github) | ✅ | ✅ | ❌ | ❌ | Provides AWS, Cohere, AI21, Antropic and Stability AI models |
64
64
  | [Cohere](https://cohere.com/?utm_source=langchainrb&utm_medium=github) | ✅ | ✅ | ✅ | ✅ | |
65
65
  | [GooglePalm](https://ai.google/discover/palm2?utm_source=langchainrb&utm_medium=github) | ✅ | ✅ | ✅ | ✅ | |
@@ -372,7 +372,7 @@ my_docx = Langchain.root.join("path/to/my.docx")
372
372
 
373
373
  client.add_data(paths: [my_pdf, my_text, my_docx])
374
374
  ```
375
- Supported file formats: docx, html, pdf, text, json, jsonl, csv, xlsx, eml.
375
+ Supported file formats: docx, html, pdf, text, json, jsonl, csv, xlsx, eml, pptx.
376
376
 
377
377
  Retrieve similar documents based on the query string passed in:
378
378
  ```ruby
@@ -129,6 +129,7 @@ module Langchain
129
129
  #
130
130
  # @return [Array] Empty messages array
131
131
  def clear_thread!
132
+ # TODO: If this a bug? Should we keep the "system" message?
132
133
  thread.messages = []
133
134
  end
134
135
 
@@ -14,12 +14,19 @@ module Langchain::LLM
14
14
  DEFAULTS = {
15
15
  temperature: 0.0,
16
16
  completion_model_name: "claude-2",
17
+ chat_completion_model_name: "claude-3-sonnet-20240229",
17
18
  max_tokens_to_sample: 256
18
19
  }.freeze
19
20
 
20
21
  # TODO: Implement token length validator for Anthropic
21
22
  # LENGTH_VALIDATOR = Langchain::Utils::TokenLength::AnthropicValidator
22
23
 
24
+ # Initialize an Anthropic LLM instance
25
+ #
26
+ # @param api_key [String] The API key to use
27
+ # @param llm_options [Hash] Options to pass to the Anthropic client
28
+ # @param default_options [Hash] Default options to use on every call to LLM, e.g.: { temperature:, completion_model_name:, chat_completion_model_name:, max_tokens_to_sample: }
29
+ # @return [Langchain::LLM::Anthropic] Langchain::LLM::Anthropic instance
23
30
  def initialize(api_key:, llm_options: {}, default_options: {})
24
31
  depends_on "anthropic"
25
32
 
@@ -27,17 +34,43 @@ module Langchain::LLM
27
34
  @defaults = DEFAULTS.merge(default_options)
28
35
  end
29
36
 
30
- #
31
37
  # Generate a completion for a given prompt
32
38
  #
33
- # @param prompt [String] The prompt to generate a completion for
34
- # @param params [Hash] extra parameters passed to Anthropic::Client#complete
39
+ # @param prompt [String] Prompt to generate a completion for
40
+ # @param model [String] The model to use
41
+ # @param max_tokens_to_sample [Integer] The maximum number of tokens to sample
42
+ # @param stop_sequences [Array<String>] The stop sequences to use
43
+ # @param temperature [Float] The temperature to use
44
+ # @param top_p [Float] The top p value to use
45
+ # @param top_k [Integer] The top k value to use
46
+ # @param metadata [Hash] The metadata to use
47
+ # @param stream [Boolean] Whether to stream the response
35
48
  # @return [Langchain::LLM::AnthropicResponse] The completion
36
- #
37
- def complete(prompt:, **params)
38
- parameters = compose_parameters @defaults[:completion_model_name], params
49
+ def complete(
50
+ prompt:,
51
+ model: @defaults[:completion_model_name],
52
+ max_tokens_to_sample: @defaults[:max_tokens_to_sample],
53
+ stop_sequences: nil,
54
+ temperature: @defaults[:temperature],
55
+ top_p: nil,
56
+ top_k: nil,
57
+ metadata: nil,
58
+ stream: nil
59
+ )
60
+ raise ArgumentError.new("model argument is required") if model.empty?
61
+ raise ArgumentError.new("max_tokens_to_sample argument is required") if max_tokens_to_sample.nil?
39
62
 
40
- parameters[:prompt] = prompt
63
+ parameters = {
64
+ model: model,
65
+ prompt: prompt,
66
+ max_tokens_to_sample: max_tokens_to_sample,
67
+ temperature: temperature
68
+ }
69
+ parameters[:stop_sequences] = stop_sequences if stop_sequences
70
+ parameters[:top_p] = top_p if top_p
71
+ parameters[:top_k] = top_k if top_k
72
+ parameters[:metadata] = metadata if metadata
73
+ parameters[:stream] = stream if stream
41
74
 
42
75
  # TODO: Implement token length validator for Anthropic
43
76
  # parameters[:max_tokens_to_sample] = validate_max_tokens(prompt, parameters[:completion_model_name])
@@ -46,12 +79,54 @@ module Langchain::LLM
46
79
  Langchain::LLM::AnthropicResponse.new(response)
47
80
  end
48
81
 
49
- private
82
+ # Generate a chat completion for given messages
83
+ #
84
+ # @param messages [Array<String>] Input messages
85
+ # @param model [String] The model that will complete your prompt
86
+ # @param max_tokens [Integer] Maximum number of tokens to generate before stopping
87
+ # @param metadata [Hash] Object describing metadata about the request
88
+ # @param stop_sequences [Array<String>] Custom text sequences that will cause the model to stop generating
89
+ # @param stream [Boolean] Whether to incrementally stream the response using server-sent events
90
+ # @param system [String] System prompt
91
+ # @param temperature [Float] Amount of randomness injected into the response
92
+ # @param tools [Array<String>] Definitions of tools that the model may use
93
+ # @param top_k [Integer] Only sample from the top K options for each subsequent token
94
+ # @param top_p [Float] Use nucleus sampling.
95
+ # @return [Langchain::LLM::AnthropicResponse] The chat completion
96
+ def chat(
97
+ messages: [],
98
+ model: @defaults[:chat_completion_model_name],
99
+ max_tokens: @defaults[:max_tokens_to_sample],
100
+ metadata: nil,
101
+ stop_sequences: nil,
102
+ stream: nil,
103
+ system: nil,
104
+ temperature: @defaults[:temperature],
105
+ tools: [],
106
+ top_k: nil,
107
+ top_p: nil
108
+ )
109
+ raise ArgumentError.new("messages argument is required") if messages.empty?
110
+ raise ArgumentError.new("model argument is required") if model.empty?
111
+ raise ArgumentError.new("max_tokens argument is required") if max_tokens.nil?
112
+
113
+ parameters = {
114
+ messages: messages,
115
+ model: model,
116
+ max_tokens: max_tokens,
117
+ temperature: temperature
118
+ }
119
+ parameters[:metadata] = metadata if metadata
120
+ parameters[:stop_sequences] = stop_sequences if stop_sequences
121
+ parameters[:stream] = stream if stream
122
+ parameters[:system] = system if system
123
+ parameters[:tools] = tools if tools.any?
124
+ parameters[:top_k] = top_k if top_k
125
+ parameters[:top_p] = top_p if top_p
50
126
 
51
- def compose_parameters(model, params)
52
- default_params = {model: model}.merge(@defaults.except(:completion_model_name))
127
+ response = client.messages(parameters: parameters)
53
128
 
54
- default_params.merge(params)
129
+ Langchain::LLM::AnthropicResponse.new(response)
55
130
  end
56
131
 
57
132
  # TODO: Implement token length validator for Anthropic
@@ -100,7 +100,7 @@ module Langchain::LLM
100
100
  end
101
101
  # rubocop:enable Style/ArgumentsForwarding
102
102
 
103
- # Generate a chat completion for a given prompt or messages.
103
+ # Generate a chat completion for given messages.
104
104
  #
105
105
  # @param messages [Array<Hash>] List of messages comprising the conversation so far
106
106
  # @param model [String] ID of the model to use
@@ -10,6 +10,10 @@ module Langchain::LLM
10
10
  completions.first
11
11
  end
12
12
 
13
+ def chat_completion
14
+ raw_response.dig("content", 0, "text")
15
+ end
16
+
13
17
  def completions
14
18
  [raw_response.dig("completion")]
15
19
  end
@@ -26,8 +30,20 @@ module Langchain::LLM
26
30
  raw_response.dig("log_id")
27
31
  end
28
32
 
33
+ def prompt_tokens
34
+ raw_response.dig("usage", "input_tokens").to_i
35
+ end
36
+
37
+ def completion_tokens
38
+ raw_response.dig("usage", "output_tokens").to_i
39
+ end
40
+
41
+ def total_tokens
42
+ prompt_tokens + completion_tokens
43
+ end
44
+
29
45
  def role
30
- "assistant"
46
+ raw_response.dig("role")
31
47
  end
32
48
  end
33
49
  end
@@ -5,15 +5,26 @@ require "csv"
5
5
  module Langchain
6
6
  module Processors
7
7
  class CSV < Base
8
+ class InvalidChunkMode < StandardError; end
9
+
8
10
  EXTENSIONS = [".csv"]
9
11
  CONTENT_TYPES = ["text/csv"]
12
+ CHUNK_MODE = {
13
+ row: "row",
14
+ file: "file"
15
+ }
10
16
 
11
17
  # Parse the document and return the text
12
18
  # @param [File] data
13
- # @return [Array of Hash]
19
+ # @return [String]
14
20
  def parse(data)
15
- ::CSV.new(data.read, col_sep: separator).map do |row|
16
- row.map(&:strip)
21
+ case chunk_mode
22
+ when CHUNK_MODE[:row]
23
+ chunk_row(data)
24
+ when CHUNK_MODE[:file]
25
+ chunk_file(data)
26
+ else
27
+ raise InvalidChunkMode
17
28
  end
18
29
  end
19
30
 
@@ -22,6 +33,29 @@ module Langchain
22
33
  def separator
23
34
  @options[:col_sep] || ","
24
35
  end
36
+
37
+ def chunk_mode
38
+ if @options[:chunk_mode].to_s.empty?
39
+ CHUNK_MODE[:row]
40
+ else
41
+ raise InvalidChunkMode unless CHUNK_MODE.value?(@options[:chunk_mode])
42
+
43
+ @options[:chunk_mode]
44
+ end
45
+ end
46
+
47
+ def chunk_row(data)
48
+ ::CSV.new(data.read, col_sep: separator).map do |row|
49
+ row
50
+ .compact
51
+ .map(&:strip)
52
+ .join(separator)
53
+ end.join("\n\n")
54
+ end
55
+
56
+ def chunk_file(data)
57
+ data.read
58
+ end
25
59
  end
26
60
  end
27
61
  end
@@ -0,0 +1,29 @@
1
+ # frozen_string_literal: true
2
+
3
+ module Langchain
4
+ module Processors
5
+ class Pptx < Base
6
+ EXTENSIONS = [".pptx"]
7
+ CONTENT_TYPES = ["application/vnd.openxmlformats-officedocument.presentationml.presentation"]
8
+
9
+ def initialize(*)
10
+ depends_on "power_point_pptx"
11
+ end
12
+
13
+ # Parse the document and return the text
14
+ # @param [File] data
15
+ # @return [String]
16
+ def parse(data)
17
+ presentation = PowerPointPptx::Document.open(data)
18
+
19
+ slides = presentation.slides
20
+ contents = slides.map(&:content)
21
+ text = contents.map do |sections|
22
+ sections.map(&:strip).join(" ")
23
+ end
24
+
25
+ text.join("\n\n")
26
+ end
27
+ end
28
+ end
29
+ end
@@ -16,7 +16,8 @@ module Langchain::Vectorsearch
16
16
  # The operators supported by the PostgreSQL vector search adapter
17
17
  OPERATORS = {
18
18
  "cosine_distance" => "cosine",
19
- "euclidean_distance" => "euclidean"
19
+ "euclidean_distance" => "euclidean",
20
+ "inner_product_distance" => "inner_product"
20
21
  }
21
22
  DEFAULT_OPERATOR = "cosine_distance"
22
23
 
@@ -1,5 +1,5 @@
1
1
  # frozen_string_literal: true
2
2
 
3
3
  module Langchain
4
- VERSION = "0.11.2"
4
+ VERSION = "0.11.3"
5
5
  end
metadata CHANGED
@@ -1,14 +1,14 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: langchainrb
3
3
  version: !ruby/object:Gem::Version
4
- version: 0.11.2
4
+ version: 0.11.3
5
5
  platform: ruby
6
6
  authors:
7
7
  - Andrei Bondarev
8
8
  autorequire:
9
9
  bindir: exe
10
10
  cert_chain: []
11
- date: 2024-04-13 00:00:00.000000000 Z
11
+ date: 2024-04-17 00:00:00.000000000 Z
12
12
  dependencies:
13
13
  - !ruby/object:Gem::Dependency
14
14
  name: activesupport
@@ -238,16 +238,16 @@ dependencies:
238
238
  name: anthropic
239
239
  requirement: !ruby/object:Gem::Requirement
240
240
  requirements:
241
- - - "~>"
241
+ - - ">="
242
242
  - !ruby/object:Gem::Version
243
- version: 0.1.0
243
+ version: '0'
244
244
  type: :development
245
245
  prerelease: false
246
246
  version_requirements: !ruby/object:Gem::Requirement
247
247
  requirements:
248
- - - "~>"
248
+ - - ">="
249
249
  - !ruby/object:Gem::Version
250
- version: 0.1.0
250
+ version: '0'
251
251
  - !ruby/object:Gem::Dependency
252
252
  name: aws-sdk-bedrockruntime
253
253
  requirement: !ruby/object:Gem::Requirement
@@ -682,6 +682,20 @@ dependencies:
682
682
  - - ">="
683
683
  - !ruby/object:Gem::Version
684
684
  version: '0'
685
+ - !ruby/object:Gem::Dependency
686
+ name: power_point_pptx
687
+ requirement: !ruby/object:Gem::Requirement
688
+ requirements:
689
+ - - "~>"
690
+ - !ruby/object:Gem::Version
691
+ version: 0.1.0
692
+ type: :development
693
+ prerelease: false
694
+ version_requirements: !ruby/object:Gem::Requirement
695
+ requirements:
696
+ - - "~>"
697
+ - !ruby/object:Gem::Version
698
+ version: 0.1.0
685
699
  description: Build LLM-backed Ruby applications with Ruby's Langchain.rb
686
700
  email:
687
701
  - andrei.bondarev13@gmail.com
@@ -758,6 +772,7 @@ files:
758
772
  - lib/langchain/processors/jsonl.rb
759
773
  - lib/langchain/processors/markdown.rb
760
774
  - lib/langchain/processors/pdf.rb
775
+ - lib/langchain/processors/pptx.rb
761
776
  - lib/langchain/processors/text.rb
762
777
  - lib/langchain/processors/xlsx.rb
763
778
  - lib/langchain/prompt.rb