langchainrb 0.6.7 → 0.6.9
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.tool-versions +1 -0
- data/CHANGELOG.md +4 -0
- data/Gemfile.lock +5 -5
- data/lib/langchain/llm/google_palm.rb +2 -2
- data/lib/langchain/llm/openai.rb +5 -2
- data/lib/langchain/utils/token_length/base_validator.rb +1 -1
- data/lib/langchain/utils/token_length/openai_validator.rb +1 -0
- data/lib/langchain/vectorsearch/pgvector.rb +48 -51
- data/lib/langchain/version.rb +1 -1
- metadata +5 -4
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 4d046a7cae545da3092694fee60b0fa3caa0852492c9bf7dbbe35001d96106e4
|
4
|
+
data.tar.gz: 8ed841d2359400f9a4700a49b82030fbe3dd3563545eca31b487914e8fb236f7
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: eed2f03cbdd74ef42fafd07524dc6bf5c7ac2278e5e55dd1aad5b2cd87f319a91c8719a6e9ac43ac50cda5525f3defab14cfecc749bdec81075d52b3935b23dd
|
7
|
+
data.tar.gz: cd4fce88b0f9f545f6d32660879e6fa5044ec74b41dcb41bc29eafb78b307d8b2cb8204856098657d8c9707331c8629c8a707a227fe4d20a5816347f8a66991c
|
data/.tool-versions
ADDED
@@ -0,0 +1 @@
|
|
1
|
+
ruby 3.1.2
|
data/CHANGELOG.md
CHANGED
data/Gemfile.lock
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
PATH
|
2
2
|
remote: .
|
3
3
|
specs:
|
4
|
-
langchainrb (0.6.
|
4
|
+
langchainrb (0.6.9)
|
5
5
|
baran (~> 0.1.6)
|
6
6
|
colorize (~> 0.8.1)
|
7
7
|
json-schema (~> 4.0.0)
|
@@ -34,7 +34,7 @@ GEM
|
|
34
34
|
afm (0.2.2)
|
35
35
|
ai21 (0.2.1)
|
36
36
|
ast (2.4.2)
|
37
|
-
baran (0.1.
|
37
|
+
baran (0.1.7)
|
38
38
|
builder (3.2.4)
|
39
39
|
byebug (11.1.3)
|
40
40
|
childprocess (4.1.0)
|
@@ -189,7 +189,7 @@ GEM
|
|
189
189
|
ruby-rc4
|
190
190
|
ttfunk
|
191
191
|
pg (1.5.3)
|
192
|
-
pgvector (0.2.
|
192
|
+
pgvector (0.2.1)
|
193
193
|
pinecone (0.1.71)
|
194
194
|
dry-struct (~> 1.6.0)
|
195
195
|
dry-validation (~> 1.10.0)
|
@@ -334,7 +334,7 @@ DEPENDENCIES
|
|
334
334
|
open-weather-ruby-client (~> 0.3.0)
|
335
335
|
pdf-reader (~> 1.4)
|
336
336
|
pg (~> 1.5)
|
337
|
-
pgvector (~> 0.2)
|
337
|
+
pgvector (~> 0.2.1)
|
338
338
|
pinecone (~> 0.1.6)
|
339
339
|
pry-byebug (~> 3.10.0)
|
340
340
|
qdrant-ruby (~> 0.9.0)
|
@@ -353,4 +353,4 @@ DEPENDENCIES
|
|
353
353
|
yard
|
354
354
|
|
355
355
|
BUNDLED WITH
|
356
|
-
2.
|
356
|
+
2.3.22
|
@@ -155,8 +155,8 @@ module Langchain::LLM
|
|
155
155
|
def transform_messages(messages)
|
156
156
|
messages.map do |message|
|
157
157
|
{
|
158
|
-
author: message[:role],
|
159
|
-
content: message[:content]
|
158
|
+
author: message[:role] || message["role"],
|
159
|
+
content: message[:content] || message["content"]
|
160
160
|
}
|
161
161
|
end
|
162
162
|
end
|
data/lib/langchain/llm/openai.rb
CHANGED
@@ -189,9 +189,12 @@ module Langchain::LLM
|
|
189
189
|
|
190
190
|
def transform_messages(messages)
|
191
191
|
messages.map do |message|
|
192
|
+
role = message[:role] || message["role"]
|
193
|
+
content = message[:content] || message["content"]
|
194
|
+
|
192
195
|
{
|
193
|
-
content:
|
194
|
-
role: (
|
196
|
+
content: content,
|
197
|
+
role: (role == "ai") ? "assistant" : role
|
195
198
|
}
|
196
199
|
end
|
197
200
|
end
|
@@ -22,7 +22,7 @@ module Langchain
|
|
22
22
|
leftover_tokens = token_limit(model_name) - text_token_length
|
23
23
|
|
24
24
|
# Raise an error even if whole prompt is equal to the model's token limit (leftover_tokens == 0)
|
25
|
-
if leftover_tokens
|
25
|
+
if leftover_tokens < 0
|
26
26
|
raise limit_exceeded_exception(token_limit(model_name), text_token_length)
|
27
27
|
end
|
28
28
|
|
@@ -8,38 +8,45 @@ module Langchain::Vectorsearch
|
|
8
8
|
# Gem requirements: gem "pgvector", "~> 0.2"
|
9
9
|
#
|
10
10
|
# Usage:
|
11
|
-
# pgvector = Langchain::Vectorsearch::Pgvector.new(url:, index_name:, llm:,
|
11
|
+
# pgvector = Langchain::Vectorsearch::Pgvector.new(url:, index_name:, llm:, namespace_column: nil, namespace: nil)
|
12
12
|
#
|
13
13
|
|
14
14
|
# The operators supported by the PostgreSQL vector search adapter
|
15
15
|
OPERATORS = {
|
16
|
-
"cosine_distance" => "
|
17
|
-
"euclidean_distance" => "
|
16
|
+
"cosine_distance" => "cosine",
|
17
|
+
"euclidean_distance" => "euclidean"
|
18
18
|
}
|
19
19
|
DEFAULT_OPERATOR = "cosine_distance"
|
20
20
|
|
21
|
-
attr_reader :operator, :
|
21
|
+
attr_reader :db, :operator, :table_name, :namespace_column, :namespace, :documents_table
|
22
22
|
|
23
23
|
# @param url [String] The URL of the PostgreSQL database
|
24
24
|
# @param index_name [String] The name of the table to use for the index
|
25
25
|
# @param llm [Object] The LLM client to use
|
26
|
-
# @param
|
27
|
-
def initialize(url:, index_name:, llm:,
|
28
|
-
|
26
|
+
# @param namespace [String] The namespace to use for the index when inserting/querying
|
27
|
+
def initialize(url:, index_name:, llm:, namespace: nil)
|
28
|
+
depends_on "sequel"
|
29
|
+
require "sequel"
|
30
|
+
depends_on "pgvector"
|
29
31
|
require "pgvector"
|
30
32
|
|
31
|
-
@
|
32
|
-
registry = ::PG::BasicTypeRegistry.new.define_default_types
|
33
|
-
::Pgvector::PG.register_vector(registry)
|
34
|
-
@client.type_map_for_results = PG::BasicTypeMapForResults.new(@client, registry: registry)
|
33
|
+
@db = Sequel.connect(url)
|
35
34
|
|
36
|
-
@
|
37
|
-
|
35
|
+
@table_name = index_name
|
36
|
+
|
37
|
+
@namespace_column = "namespace"
|
38
|
+
@namespace = namespace
|
38
39
|
@operator = OPERATORS[DEFAULT_OPERATOR]
|
39
40
|
|
40
41
|
super(llm: llm)
|
41
42
|
end
|
42
43
|
|
44
|
+
def documents_model
|
45
|
+
Class.new(Sequel::Model(table_name.to_sym)) do
|
46
|
+
plugin :pgvector, :vectors
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
43
50
|
# Upsert a list of texts to the index
|
44
51
|
# @param texts [Array<String>] The texts to add to the index
|
45
52
|
# @param ids [Array<Integer>] The ids of the objects to add to the index, in the same order as the texts
|
@@ -47,32 +54,28 @@ module Langchain::Vectorsearch
|
|
47
54
|
# the added or updated texts.
|
48
55
|
def upsert_texts(texts:, ids:)
|
49
56
|
data = texts.zip(ids).flat_map do |(text, id)|
|
50
|
-
|
57
|
+
{id: id, content: text, vectors: llm.embed(text: text).to_s, namespace: namespace}
|
51
58
|
end
|
52
|
-
|
53
|
-
|
54
|
-
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
+
# @db[table_name.to_sym].multi_insert(data, return: :primary_key)
|
60
|
+
@db[table_name.to_sym]
|
61
|
+
.insert_conflict(
|
62
|
+
target: :id,
|
63
|
+
update: {content: Sequel[:excluded][:content], vectors: Sequel[:excluded][:vectors]}
|
64
|
+
)
|
65
|
+
.multi_insert(data, return: :primary_key)
|
59
66
|
end
|
60
67
|
|
61
68
|
# Add a list of texts to the index
|
62
69
|
# @param texts [Array<String>] The texts to add to the index
|
63
70
|
# @param ids [Array<String>] The ids to add to the index, in the same order as the texts
|
64
|
-
# @return [
|
65
|
-
# the added texts.
|
71
|
+
# @return [Array<Integer>] The the ids of the added texts.
|
66
72
|
def add_texts(texts:, ids: nil)
|
67
73
|
if ids.nil? || ids.empty?
|
68
|
-
data = texts.
|
69
|
-
|
74
|
+
data = texts.map do |text|
|
75
|
+
{content: text, vectors: llm.embed(text: text).to_s, namespace: namespace}
|
70
76
|
end
|
71
|
-
|
72
|
-
|
73
|
-
"INSERT INTO #{quoted_table_name} (content, vectors) VALUES #{values} RETURNING id;",
|
74
|
-
data
|
75
|
-
)
|
77
|
+
|
78
|
+
@db[table_name.to_sym].multi_insert(data, return: :primary_key)
|
76
79
|
else
|
77
80
|
upsert_texts(texts: texts, ids: ids)
|
78
81
|
end
|
@@ -81,8 +84,7 @@ module Langchain::Vectorsearch
|
|
81
84
|
# Update a list of ids and corresponding texts to the index
|
82
85
|
# @param texts [Array<String>] The texts to add to the index
|
83
86
|
# @param ids [Array<String>] The ids to add to the index, in the same order as the texts
|
84
|
-
# @return [
|
85
|
-
# the updated texts.
|
87
|
+
# @return [Array<Integer>] The ids of the updated texts.
|
86
88
|
def update_texts(texts:, ids:)
|
87
89
|
upsert_texts(texts: texts, ids: ids)
|
88
90
|
end
|
@@ -90,16 +92,15 @@ module Langchain::Vectorsearch
|
|
90
92
|
# Create default schema
|
91
93
|
# @return [PG::Result] The response from the database
|
92
94
|
def create_default_schema
|
93
|
-
|
94
|
-
|
95
|
-
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
|
100
|
-
|
101
|
-
|
102
|
-
)
|
95
|
+
db.run "CREATE EXTENSION IF NOT EXISTS vector"
|
96
|
+
namespace = namespace_column
|
97
|
+
vector_dimension = default_dimension
|
98
|
+
db.create_table? table_name.to_sym do
|
99
|
+
primary_key :id
|
100
|
+
text :content
|
101
|
+
column :vectors, "vector(#{vector_dimension})"
|
102
|
+
text namespace.to_sym, default: nil
|
103
|
+
end
|
103
104
|
end
|
104
105
|
|
105
106
|
# TODO: Add destroy_default_schema method
|
@@ -123,15 +124,11 @@ module Langchain::Vectorsearch
|
|
123
124
|
# @param k [Integer] The number of top results to return
|
124
125
|
# @return [Array<Hash>] The results of the search
|
125
126
|
def similarity_search_by_vector(embedding:, k: 4)
|
126
|
-
|
127
|
-
|
128
|
-
|
129
|
-
|
130
|
-
SQL
|
131
|
-
conn.exec_params(query, [embedding, k])
|
127
|
+
db.transaction do # BEGIN
|
128
|
+
documents_model
|
129
|
+
.nearest_neighbors(:vectors, embedding, distance: operator).limit(k)
|
130
|
+
.where(namespace_column.to_sym => namespace)
|
132
131
|
end
|
133
|
-
|
134
|
-
result.to_a
|
135
132
|
end
|
136
133
|
|
137
134
|
# Ask a question and return the answer
|
@@ -142,7 +139,7 @@ module Langchain::Vectorsearch
|
|
142
139
|
search_results = similarity_search(query: question)
|
143
140
|
|
144
141
|
context = search_results.map do |result|
|
145
|
-
result
|
142
|
+
result.content.to_s
|
146
143
|
end
|
147
144
|
context = context.join("\n---\n")
|
148
145
|
|
data/lib/langchain/version.rb
CHANGED
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: 0.6.
|
4
|
+
version: 0.6.9
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Andrei Bondarev
|
8
8
|
autorequire:
|
9
9
|
bindir: exe
|
10
10
|
cert_chain: []
|
11
|
-
date: 2023-07-
|
11
|
+
date: 2023-07-29 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: baran
|
@@ -324,14 +324,14 @@ dependencies:
|
|
324
324
|
requirements:
|
325
325
|
- - "~>"
|
326
326
|
- !ruby/object:Gem::Version
|
327
|
-
version:
|
327
|
+
version: 0.2.1
|
328
328
|
type: :development
|
329
329
|
prerelease: false
|
330
330
|
version_requirements: !ruby/object:Gem::Requirement
|
331
331
|
requirements:
|
332
332
|
- - "~>"
|
333
333
|
- !ruby/object:Gem::Version
|
334
|
-
version:
|
334
|
+
version: 0.2.1
|
335
335
|
- !ruby/object:Gem::Dependency
|
336
336
|
name: pdf-reader
|
337
337
|
requirement: !ruby/object:Gem::Requirement
|
@@ -482,6 +482,7 @@ files:
|
|
482
482
|
- ".env.example"
|
483
483
|
- ".rspec"
|
484
484
|
- ".rubocop.yml"
|
485
|
+
- ".tool-versions"
|
485
486
|
- CHANGELOG.md
|
486
487
|
- Gemfile
|
487
488
|
- Gemfile.lock
|