j1-template 2024.1.5 → 2024.2.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/_includes/themes/j1/layouts/content_generator_page.html +0 -34
- data/_includes/themes/j1/modules/searcher/generator.html +4 -1
- data/_includes/themes/j1/modules/searcher/procedures/topsearch.proc +59 -56
- data/assets/data/authclient.html +1 -8
- data/assets/data/banner.html +1 -7
- data/assets/data/fab.html +2 -13
- data/assets/data/footer.html +0 -3
- data/assets/data/galeries.html +2 -13
- data/assets/data/gemini-ui.html +74 -21
- data/assets/data/iframes.html +1 -4
- data/assets/data/masonry.html +1 -13
- data/assets/data/masterslider.html +10 -21
- data/assets/data/menu.html +6 -14
- data/assets/data/mmenu.html +6 -5
- data/assets/data/mmenu_toc.html +2 -1
- data/assets/data/panel.html +4 -11
- data/assets/data/particles.yml +8 -8
- data/assets/data/quicklinks.html +4 -23
- data/assets/data/rtext_resizer.html +3 -1
- data/assets/data/slick.html +9 -11
- data/assets/data/translator.html +0 -1
- data/assets/themes/j1/adapter/js/advertising.js +143 -182
- data/assets/themes/j1/adapter/js/algolia.js +61 -54
- data/assets/themes/j1/adapter/js/analytics.js +68 -52
- data/assets/themes/j1/adapter/js/asciidoctor.js +32 -20
- data/assets/themes/j1/adapter/js/attic.js +75 -69
- data/assets/themes/j1/adapter/js/bmd.js +44 -26
- data/assets/themes/j1/adapter/js/carousel.js +117 -92
- data/assets/themes/j1/adapter/js/chatbot.js +77 -35
- data/assets/themes/j1/adapter/js/clipboard.js +67 -49
- data/assets/themes/j1/adapter/js/comments.js +93 -70
- data/assets/themes/j1/adapter/js/cookieConsent.js +157 -159
- data/assets/themes/j1/adapter/js/customFunctions.js +52 -35
- data/assets/themes/j1/adapter/js/customModule.js +61 -50
- data/assets/themes/j1/adapter/js/docsearch.js +55 -34
- data/assets/themes/j1/adapter/js/dropdowns.js +66 -52
- data/assets/themes/j1/adapter/js/fab.js +124 -109
- data/assets/themes/j1/adapter/js/gallery.js +68 -49
- data/assets/themes/j1/adapter/js/gemini.js +961 -301
- data/assets/themes/j1/adapter/js/iconPicker.js +76 -54
- data/assets/themes/j1/adapter/js/iconPickerPage.js +279 -0
- data/assets/themes/j1/adapter/js/iframer.js +84 -63
- data/assets/themes/j1/adapter/js/j1.js +605 -542
- data/assets/themes/j1/adapter/js/lazyLoader.js +43 -24
- data/assets/themes/j1/adapter/js/lightbox.js +53 -52
- data/assets/themes/j1/adapter/js/logger.js +78 -66
- data/assets/themes/j1/adapter/js/lunr.js +684 -86
- data/assets/themes/j1/adapter/js/masonry.js +84 -68
- data/assets/themes/j1/adapter/js/masterslider.js +136 -137
- data/assets/themes/j1/adapter/js/mmenu.js +102 -66
- data/assets/themes/j1/adapter/js/navigator.js +291 -356
- data/assets/themes/j1/adapter/js/particles.js +62 -40
- data/assets/themes/j1/adapter/js/rangeSlider.js +66 -48
- data/assets/themes/j1/adapter/js/rouge.js +63 -44
- data/assets/themes/j1/adapter/js/rtable.js +56 -40
- data/assets/themes/j1/adapter/js/rtextResizer.js +57 -44
- data/assets/themes/j1/adapter/js/scroller.js +170 -133
- data/assets/themes/j1/adapter/js/slick.js +99 -97
- data/assets/themes/j1/adapter/js/slimSelect.js +288 -0
- data/assets/themes/j1/adapter/js/speak2me.js +124 -128
- data/assets/themes/j1/adapter/js/themeToggler.js +118 -97
- data/assets/themes/j1/adapter/js/{themer.js → themes.js} +144 -113
- data/assets/themes/j1/adapter/js/toccer.js +88 -54
- data/assets/themes/j1/adapter/js/translator.js +74 -70
- data/assets/themes/j1/adapter/js/waves.js +75 -58
- data/assets/themes/j1/core/css/themes/unolight/bootstrap.css +29 -11
- data/assets/themes/j1/core/css/themes/unolight/bootstrap.min.css +2 -2
- data/assets/themes/j1/core/js/template.js +474 -420
- data/assets/themes/j1/core/js/template.min.js +7 -7
- data/assets/themes/j1/core/js/template.min.js.map +1 -1
- data/assets/themes/j1/modules/carousel/css/theme/uno.css +4 -4
- data/assets/themes/j1/modules/carousel/css/theme/uno.min.css +1 -1
- data/assets/themes/j1/modules/clipboard/js/clipboard.js +0 -1
- data/assets/themes/j1/modules/cookieConsent/js/cookieConsent.js +27 -21
- data/assets/themes/j1/modules/iconPicker/css/theme/uno.css +58 -0
- data/assets/themes/j1/modules/iconPicker/css/theme/uno.min.css +16 -0
- data/assets/themes/j1/modules/iconPicker/js/universal-icon-picker.0.js +493 -0
- data/assets/themes/j1/modules/iconPicker/js/universal-icon-picker.js +7 -7
- data/assets/themes/j1/modules/js-cookies/js/js.cookie.js +147 -0
- data/assets/themes/j1/modules/js-cookies/js/js.cookie.min.js +2 -0
- data/assets/themes/j1/modules/lightGallery/js/plugins/lg-video.js +4 -4
- data/assets/themes/j1/modules/lightGallery/js/plugins/lg-video.min.js +2 -10
- data/assets/themes/j1/modules/lunr/css/j1.css +12 -6
- data/assets/themes/j1/modules/lunr/css/j1.min.css +1 -1
- data/assets/themes/j1/modules/lunr/js/j1.js +46 -43
- data/assets/themes/j1/modules/lunr/js/j1.min.js +1 -1
- data/assets/themes/j1/modules/masterslider/js/masterslider.js +1 -1
- data/assets/themes/j1/modules/scroller/js/scroller.js +64 -74
- data/assets/themes/j1/modules/scroller/js/scroller.min.js +1 -1
- data/assets/themes/j1/modules/slick/slider/css/theme/uno.css +4 -4
- data/assets/themes/j1/modules/slick/slider/css/theme/uno.min.css +1 -1
- data/assets/themes/j1/modules/slimSelect/js/select.js +1865 -1821
- data/assets/themes/j1/modules/slimSelect/js/select.min.js +2 -1
- data/assets/themes/j1/modules/themeSwitcher/js/switcher.js +87 -89
- data/assets/themes/j1/modules/themeSwitcher/js/switcher.min.js +1 -1
- data/assets/themes/j1/modules/videojs/js/plugins/vm/api/player.min.js +5 -7
- data/assets/themes/j1/modules/videojs/js/plugins/vm/api/v2.20.1/player.min.js +23 -0
- data/lib/j1/version.rb +1 -1
- data/lib/starter_web/Gemfile +2 -2
- data/lib/starter_web/README.md +5 -5
- data/lib/starter_web/_config.yml +1 -1
- data/lib/starter_web/_data/blocks/_panel.yml +775 -0
- data/lib/starter_web/_data/blocks/panel.yml +53 -53
- data/lib/starter_web/_data/j1_config.yml +3 -2
- data/lib/starter_web/_data/layouts/default.yml +0 -2
- data/lib/starter_web/_data/modules/_scroller.yml +102 -0
- data/lib/starter_web/_data/modules/carousel.yml +3 -3
- data/lib/starter_web/_data/modules/defaults/attics.yml +5 -5
- data/lib/starter_web/_data/modules/defaults/docsearch.yml +1 -1
- data/lib/starter_web/_data/modules/defaults/gemini.yml +204 -46
- data/lib/starter_web/_data/modules/defaults/{iconPicker.yml → icon_picker.yml} +6 -12
- data/lib/starter_web/_data/modules/defaults/lunr.yml +20 -5
- data/lib/starter_web/_data/modules/defaults/masterslider.yml +4 -4
- data/lib/starter_web/_data/modules/defaults/navigator.yml +20 -24
- data/lib/starter_web/_data/modules/defaults/particles.yml +3 -3
- data/lib/starter_web/_data/modules/defaults/slim_select.yml +54 -0
- data/lib/starter_web/_data/modules/defaults/{themer.yml → themes.yml} +10 -10
- data/lib/starter_web/_data/modules/defaults/toccer.yml +1 -1
- data/lib/starter_web/_data/modules/gallery.yml +33 -38
- data/lib/starter_web/_data/modules/gemini.yml +42 -3
- data/lib/starter_web/_data/modules/{iconPicker.yml → icon_picker.yml} +31 -3
- data/lib/starter_web/_data/modules/lunr.yml +12 -1
- data/lib/starter_web/_data/modules/masonry.yml +37 -38
- data/lib/starter_web/_data/modules/masterslider.yml +78 -95
- data/lib/starter_web/_data/modules/navigator_menu.yml +12 -20
- data/lib/starter_web/_data/modules/particles.yml +3 -3
- data/lib/starter_web/_data/modules/scroller.yml +3 -3
- data/lib/starter_web/_data/modules/slim_select.yml +110 -0
- data/lib/starter_web/_data/modules/{themer.yml → themes.yml} +4 -4
- data/lib/starter_web/_data/resources.yml +83 -47
- data/lib/starter_web/_data/templates/feed.xml +1 -1
- data/lib/starter_web/_includes/attributes.asciidoc +354 -355
- data/lib/starter_web/_plugins/asciidoctor/gemini-ui-block.rb +1 -1
- data/lib/starter_web/_plugins/asciidoctor/slim-select-block.rb +45 -0
- data/lib/starter_web/_plugins/index/lunr.rb +1 -1
- data/lib/starter_web/collections/asciidoc_skeletons/simple-post/_posts/yyyy-mm-dd-your-post-name.asciidoc +5 -2
- data/lib/starter_web/collections/posts/public/featured/_posts/0000-00-00-welcome-to-j1.adoc.erb +3 -3
- data/lib/starter_web/collections/posts/public/featured/_posts/2021-01-01-about-cookies.adoc +9 -9
- data/lib/starter_web/collections/posts/public/featured/_posts/2021-02-01-static-site-generators.adoc +4 -4
- data/lib/starter_web/collections/posts/public/featured/_posts/2022-02-01-about-j1.adoc +3 -2
- data/lib/starter_web/collections/posts/public/featured/_posts/2023-10-18-url-types.adoc +12 -12
- data/lib/starter_web/package.json +1 -1
- data/lib/starter_web/pages/public/about/features.adoc +1 -1
- data/lib/starter_web/pages/public/about/reporting_issues.adoc +1 -0
- data/lib/starter_web/pages/public/asciidoc_skeletons/documentation/_includes/attributes.asciidoc +44 -44
- data/lib/starter_web/pages/public/asciidoc_skeletons/documentation/documentation.adoc +1 -0
- data/lib/starter_web/pages/public/asciidoc_skeletons/multi-document/_includes/attributes.asciidoc +12 -12
- data/lib/starter_web/pages/public/blog/navigator/archive/allview.html +1 -1
- data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/attributes.asciidoc +42 -28
- data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/highlghter_rouge.adoc +1 -0
- data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/lunr_search.adoc +1 -0
- data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/present_audio_video.adoc +18 -2
- data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/present_images.adoc +43 -50
- data/lib/starter_web/pages/public/legal/en/100_copyright.adoc +6 -11
- data/lib/starter_web/pages/public/legal/en/200_impress.adoc +4 -11
- data/lib/starter_web/pages/public/legal/en/300_privacy.adoc +5 -12
- data/lib/starter_web/pages/public/legal/en/400_comment_policy.adoc +5 -10
- data/lib/starter_web/pages/public/tools/previewer/_includes/attributes.asciidoc +10 -9
- data/lib/starter_web/pages/public/tools/previewer/preview_bootstrap_theme.adoc +77 -75
- metadata +52 -46
- data/lib/starter_web/pages/public/manuals/integrations/gemini/_includes/attributes.asciidoc +0 -47
- data/lib/starter_web/pages/public/manuals/integrations/gemini/_includes/documents/preview_google_adsense.asciidoc +0 -448
- data/lib/starter_web/pages/public/manuals/integrations/gemini/_includes/documents/readme +0 -0
- data/lib/starter_web/pages/public/manuals/integrations/gemini/_includes/tables/readme +0 -0
- data/lib/starter_web/pages/public/manuals/integrations/gemini/gemini.adoc +0 -525
- data/lib/starter_web/pages/public/manuals/integrations/gemini/security.asccidoc +0 -274
- data/lib/starter_web/pages/public/manuals/integrations/gemini/security.hrml +0 -560
- /data/lib/starter_web/pages/public/{legal/learn → learn}/bookshelf/article_previewer/viewer_biography.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/bookshelf/article_previewer/viewer_fantasy.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/bookshelf/article_previewer/viewer_romance.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/bookshelf/jekyll_collections.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/bookshelf/viewer_all_books.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/100_gistblock.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_bottom_info.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_bottom_left_warning.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_bottom_right_danger.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_central_success.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_full_height_left_info.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_full_height_right_success.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_table_bs_modal_examples.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_top_info.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_top_left_info.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/410_top_right_success.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/419_advanced_modals_demo.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/tables/bs_modal_examples.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/themes_bootstrap.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/_includes/documents/themes_rouge.asciidoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/asciidoc_extensions.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/bootstrap_themes.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/icon_fonts.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/modal_extentions.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/responsive_tables.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/roundtrip/typography.adoc +0 -0
- /data/lib/starter_web/pages/public/{legal/learn → learn}/where_to_go.adoc +0 -0
@@ -1,525 +0,0 @@
|
|
1
|
-
---
|
2
|
-
title: Google Gemini
|
3
|
-
title_extention: Gemini API Integration for J1 Template
|
4
|
-
tagline: User Guide for J1 Template
|
5
|
-
|
6
|
-
date: 2022-08-14
|
7
|
-
|
8
|
-
description: >
|
9
|
-
Google Gemini refers to a family of large language models (LLMs)
|
10
|
-
developed by Google DeepMind. It was announced on December 6, 2023,
|
11
|
-
and is seen as the successor to "LaMDA" and "PaLM 2".
|
12
|
-
keywords: >
|
13
|
-
Jekyll, Theme, JekyllOne, Info, Google,
|
14
|
-
Gemini, Manual
|
15
|
-
|
16
|
-
categories: [ Manual, Module ]
|
17
|
-
tags: [ Google, Gemini ]
|
18
|
-
|
19
|
-
image:
|
20
|
-
path: /assets/images/modules/attics/1920x1280/build-with-gemini.jpg
|
21
|
-
width: 1920
|
22
|
-
height: 1280
|
23
|
-
|
24
|
-
tts: true
|
25
|
-
comments: false
|
26
|
-
exclude_from_search: false
|
27
|
-
regenerate: false
|
28
|
-
permalink: /pages/public/manuals/integrations/google/gemini/
|
29
|
-
|
30
|
-
resources: [
|
31
|
-
animate,
|
32
|
-
clipboard, rouge
|
33
|
-
]
|
34
|
-
resource_options:
|
35
|
-
- attic:
|
36
|
-
slides:
|
37
|
-
- url: /assets/images/modules/attics/1920x1280/build-with-gemini.jpg
|
38
|
-
alt: Build with Gemini
|
39
|
-
# badge:
|
40
|
-
# type: unsplash
|
41
|
-
# author: Eleni Afiontzi
|
42
|
-
# href: //unsplash.com/de/@eleniafiontzi
|
43
|
-
---
|
44
|
-
|
45
|
-
// Page Initializer
|
46
|
-
// =============================================================================
|
47
|
-
// Enable the Liquid Preprocessor
|
48
|
-
:page-liquid:
|
49
|
-
|
50
|
-
// Set (local) page attributes here
|
51
|
-
// -----------------------------------------------------------------------------
|
52
|
-
// :page--attr: <attr-value>
|
53
|
-
:images-dir: {imagesdir}/pages/roundtrip/100_present_images
|
54
|
-
|
55
|
-
// Load Liquid procedures
|
56
|
-
// -----------------------------------------------------------------------------
|
57
|
-
{% capture load_attributes %}themes/{{site.template.name}}/procedures/global/attributes_loader.proc{%endcapture%}
|
58
|
-
|
59
|
-
// Load page attributes
|
60
|
-
// -----------------------------------------------------------------------------
|
61
|
-
{% include {{load_attributes}} scope="all" %}
|
62
|
-
|
63
|
-
|
64
|
-
// Page content
|
65
|
-
// ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|
66
|
-
|
67
|
-
// Include sub-documents (if any)
|
68
|
-
// -----------------------------------------------------------------------------
|
69
|
-
[role="dropcap"]
|
70
|
-
Google Gemini refers to a family of large language models (LLMs) developed by
|
71
|
-
Google DeepMind. It was announced on December 6, 2023, and is seen as the
|
72
|
-
successor to LaMDA and PaLM 2.
|
73
|
-
|
74
|
-
The Gemini family consists of three models:
|
75
|
-
|
76
|
-
Gemini Ultra::
|
77
|
-
The largest and most capable model. It excels at multimodal tasks, which
|
78
|
-
involve understanding and generating text, code, and other forms of data.
|
79
|
-
|
80
|
-
Gemini Pro::
|
81
|
-
A smaller model, designed for a balance of performance and efficiency. It
|
82
|
-
is well-suited for tasks such as question answering, summarization, and
|
83
|
-
translation.
|
84
|
-
|
85
|
-
Gemini Nano::
|
86
|
-
The smallest model, optimized for mobile and edge devices. It can still
|
87
|
-
perform basic language tasks such as text generation and translation.
|
88
|
-
|
89
|
-
|
90
|
-
Gemini is still under development, but it has already achieved impressive
|
91
|
-
results on a range of benchmarks. For example, Gemini Ultra achieved a
|
92
|
-
state-of-the-art score of 59.4% on the new MMMU benchmark, which consists
|
93
|
-
of multimodal tasks spanning different domains requiring deliberate reasoning.
|
94
|
-
|
95
|
-
Google is making Gemini models available through Google AI Studio and
|
96
|
-
Google Cloud Vertex AI, so developers can integrate them into their own
|
97
|
-
applications. The company is also releasing a limited number of Gemini 1.5
|
98
|
-
Pro models for early testing.
|
99
|
-
|
100
|
-
[role="mt-5"]
|
101
|
-
== Gemini Pro
|
102
|
-
|
103
|
-
The first version of Gemini Pro is now accessible via the Gemini API and
|
104
|
-
here’s more about it:
|
105
|
-
|
106
|
-
* Gemini Pro outperforms other similarly-sized models on research
|
107
|
-
benchmarks.
|
108
|
-
* Today’s version comes with a 32K context window for text, and future
|
109
|
-
versions will have a larger context window.
|
110
|
-
* It’s free to use right now, within limits, and it will be competitively
|
111
|
-
priced.
|
112
|
-
* It comes with a range of features: function calling, embeddings, semantic
|
113
|
-
retrieval and custom knowledge grounding, and chat functionality.
|
114
|
-
* It supports 38 languages across 180+ countries and territories worldwide.
|
115
|
-
|
116
|
-
In today’s release, Gemini Pro accepts text as input and generates text as##
|
117
|
-
output. It is also made a dedicated Gemini Pro Vision multimodal endpoint
|
118
|
-
available today that accepts text and imagery as input, with text output.
|
119
|
-
|
120
|
-
SDKs are available for Gemini Pro to help you build apps that run anywhere.
|
121
|
-
Python, Android (Kotlin), Node.js, Swift and JavaScript are all supported.
|
122
|
-
|
123
|
-
|
124
|
-
[role="mt-5"]
|
125
|
-
== Google AI Studio
|
126
|
-
// https://ai.google.dev/docs
|
127
|
-
|
128
|
-
link://ai.google.dev/[Google AI Studio, {browser-window--new}] is a free,
|
129
|
-
web-based developer tool that enables you to quickly develop prompts and
|
130
|
-
then get an API key to use in your app development.
|
131
|
-
|
132
|
-
You can sign into Google AI Studio with your Google account and take
|
133
|
-
advantage of the free quota, which allows 60 requests per minute; 20x
|
134
|
-
more than other free offerings. When you’re ready, you can simply click
|
135
|
-
on “Get code” to transfer your work to your IDE of choice, or use one of
|
136
|
-
the quickstart templates available in Android Studio, Colab or Project IDX.
|
137
|
-
|
138
|
-
To help us improve product quality, when you use the free quota, your API
|
139
|
-
and Google AI Studio input and output may be accessible to trained reviewers.
|
140
|
-
This data is de-identified from your Google account and API key.
|
141
|
-
|
142
|
-
|
143
|
-
[role="mt-5"]
|
144
|
-
== Safety settings
|
145
|
-
|
146
|
-
This guide describes the adjustable safety settings available for the
|
147
|
-
Gemini API. During the prototyping stage, you can adjust safety settings
|
148
|
-
on 4 dimensions to quickly assess if your application requires more or
|
149
|
-
less restrictive configuration.
|
150
|
-
|
151
|
-
By default, safety settings block
|
152
|
-
content (including prompts) with medium or higher probability of being
|
153
|
-
unsafe across any dimension. This baseline safety is designed to work
|
154
|
-
for most use cases, so you should only adjust your safety settings if
|
155
|
-
it's consistently required for your application.
|
156
|
-
|
157
|
-
[NOTE]
|
158
|
-
====
|
159
|
-
Adjusting to lower safety settings will trigger a more indepth
|
160
|
-
review process of your application.
|
161
|
-
====
|
162
|
-
|
163
|
-
[role="mt-4"]
|
164
|
-
=== Safety filters
|
165
|
-
|
166
|
-
In addition to the adjustable safety filters, the Gemini API has
|
167
|
-
built-in protections against core harms, such as content that endangers
|
168
|
-
child safety. These types of harm are always blocked and cannot be
|
169
|
-
adjusted.
|
170
|
-
|
171
|
-
The adjustable safety filters cover the following categories:
|
172
|
-
|
173
|
-
* Harassment
|
174
|
-
* Hate speech
|
175
|
-
* Sexually explicit
|
176
|
-
* Dangerous
|
177
|
-
|
178
|
-
These settings allow the developer to determine what is appropriate for
|
179
|
-
your use case. For example, if you're building a video game dialogue,
|
180
|
-
you may deem it acceptable to allow more content rated as dangerous due
|
181
|
-
to the nature of the game. Here are a few other example use cases that
|
182
|
-
may need some flexibility in these safety settings:
|
183
|
-
|
184
|
-
[cols="4a, 8a", options="header", width="100%", role="rtable mt-3 mb-5"]
|
185
|
-
|===
|
186
|
-
|Use Case |Category
|
187
|
-
|
188
|
-
|Anti-Harassment Training App
|
189
|
-
|Hate speech, Sexually explicit
|
190
|
-
|
191
|
-
|Screenplay Writer
|
192
|
-
|Sexually explicit, Dangerous
|
193
|
-
|
194
|
-
|Toxicity classifier
|
195
|
-
|Harassment, Dangerous
|
196
|
-
|
197
|
-
|===
|
198
|
-
|
199
|
-
[role="mt-4"]
|
200
|
-
==== Probability versus severity
|
201
|
-
|
202
|
-
The Gemini API blocks content based on the *probability* of content
|
203
|
-
being unsafe and *not* the severity. This is important to consider
|
204
|
-
because some content can have low probability of being unsafe even
|
205
|
-
though the severity of harm could still be high. For example, comparing
|
206
|
-
the sentences:
|
207
|
-
|
208
|
-
. The robot punched me.
|
209
|
-
. The robot slashed me up.
|
210
|
-
|
211
|
-
Sentence 1 might result in a higher probability of being unsafe but you
|
212
|
-
might consider sentence 2 to be a higher severity in terms of violence.
|
213
|
-
|
214
|
-
Given this, it is important for each developer to carefully test and
|
215
|
-
consider what the appropriate level of blocking is needed to support
|
216
|
-
their key use cases while minimizing harm to end users.
|
217
|
-
|
218
|
-
[role="mt-4"]
|
219
|
-
==== Safety Settings
|
220
|
-
|
221
|
-
Safety settings are part of the request you send to the text service. It
|
222
|
-
can be adjusted for each request you make to the API. The following
|
223
|
-
table lists the categories that you can set and describes the type of
|
224
|
-
harm that each category encompasses.
|
225
|
-
|
226
|
-
[cols="4a, 8a", options="header", width="100%", role="rtable mt-3 mb-5"]
|
227
|
-
|===
|
228
|
-
|Categories |Descriptions
|
229
|
-
|
230
|
-
|Harassment
|
231
|
-
|Negative or harmful comments targeting identity and/or
|
232
|
-
protected attributes.
|
233
|
-
|
234
|
-
|Hate speech
|
235
|
-
|Content that is rude, disrespectful, or profane.
|
236
|
-
|
237
|
-
|Sexually explicit
|
238
|
-
|Contains references to sexual acts or other lewd
|
239
|
-
content.
|
240
|
-
|
241
|
-
|Dangerous
|
242
|
-
|Promotes, facilitates, or encourages harmful acts.
|
243
|
-
|
244
|
-
|===
|
245
|
-
|
246
|
-
These definitions are in the
|
247
|
-
link://ai.google.dev/api/rest/v1/HarmCategory[API reference, {browser-window--new}]
|
248
|
-
as well. The Gemini models only support `HARM_CATEGORY_HARASSMENT`,
|
249
|
-
`HARM_CATEGORY_HATE_SPEECH`, `HARM_CATEGORY_SEXUALLY_EXPLICIT`, and
|
250
|
-
`HARM_CATEGORY_DANGEROUS_CONTENT`.
|
251
|
-
|
252
|
-
The following table describes the block settings you can adjust for each
|
253
|
-
category. For example, if you set the block setting to *Block few* for
|
254
|
-
the *Hate speech* category, everything that has a high probability of
|
255
|
-
being hate speech content is blocked. But anything with a lower
|
256
|
-
probability is allowed.
|
257
|
-
|
258
|
-
If not set, the default block setting is *Block some* for all
|
259
|
-
categories.
|
260
|
-
|
261
|
-
[cols="4a, 8a", options="header", width="100%", role="rtable mt-3 mb-5"]
|
262
|
-
|===
|
263
|
-
|Threshold (API) |Description
|
264
|
-
|
265
|
-
|BLOCK_NONE
|
266
|
-
|Always show regardless of probability of unsafe content
|
267
|
-
|
268
|
-
|BLOCK_ONLY_HIGH
|
269
|
-
|Block when high probability of unsafe content
|
270
|
-
|
271
|
-
|BLOCK_MEDIUM_AND_ABOVE
|
272
|
-
|Block when medium or high probability of unsafe content
|
273
|
-
|
274
|
-
|BLOCK_LOW_AND_ABOVE
|
275
|
-
|Block when low, medium or high
|
276
|
-
probability of unsafe content
|
277
|
-
|
278
|
-
|HARM_BLOCK_THRESHOLD_UNSPECIFIED
|
279
|
-
|Threshold is unspecified, block using default threshold
|
280
|
-
|
281
|
-
|===
|
282
|
-
|
283
|
-
You can set these settings for each request you make to the text
|
284
|
-
service. See the
|
285
|
-
link://ai.google.dev/api/rest/v1/SafetySetting#harmblockthreshold[HarmBlockThreshold, {browser-window--new}]
|
286
|
-
API reference for details.
|
287
|
-
|
288
|
-
[role="mt-4"]
|
289
|
-
==== Safety feedback
|
290
|
-
|
291
|
-
link://ai.google.dev/api/rest/v1/models/generateContent[generateContent, {browser-window--new}]
|
292
|
-
returns a
|
293
|
-
link //ai.google.dev/api/rest/v1/GenerateContentResponse[GenerateContentResponse, {browser-window--new}]
|
294
|
-
which includes safety feedback.
|
295
|
-
|
296
|
-
Prompt feedback is included in
|
297
|
-
link://ai.google.dev/api/rest/v1/GenerateContentResponse#PromptFeedback[promptFeedback, {browser-window--new}].
|
298
|
-
If
|
299
|
-
link://ai.google.dev/api/rest/v1/GenerateContentResponse#blockreason[promptFeedback.blockReason, {browser-window--new}]
|
300
|
-
is set, then the content of the prompt was blocked.
|
301
|
-
|
302
|
-
Response candidate feedback is included in
|
303
|
-
link://ai.google.dev/api/rest/v1/GenerateContentResponse#FinishReason[finishReason, {browser-window--new}]
|
304
|
-
and
|
305
|
-
link://ai.google.dev/api/rest/v1/GenerateContentResponse#safetyrating[safetyRatings, {browser-window--new}].
|
306
|
-
|
307
|
-
If response content was blocked and the `finishReason` was `SAFETY`, you
|
308
|
-
can inspect `safetyRatings` for more details. The safety rating includes
|
309
|
-
the category and the probability of the harm classification. The content
|
310
|
-
that was blocked is not returned.
|
311
|
-
|
312
|
-
The probability returned correspond to the block confidence levels as
|
313
|
-
shown in the following table:
|
314
|
-
|
315
|
-
[cols="4a, 8a", options="header", width="100%", role="rtable mt-3 mb-5"]
|
316
|
-
|===
|
317
|
-
|Probability |Description
|
318
|
-
|
319
|
-
|NEGLIGIBLE
|
320
|
-
|Content has a negligible probability of being unsafe
|
321
|
-
|
322
|
-
|LOW
|
323
|
-
|Content has a low probability of being unsafe
|
324
|
-
|
325
|
-
|MEDIUM
|
326
|
-
|Content has a medium probability of being unsafe
|
327
|
-
|
328
|
-
|HIGH
|
329
|
-
|Content has a high probability of being unsafe
|
330
|
-
|
331
|
-
|===
|
332
|
-
|
333
|
-
For example, if the content was blocked due to the harassment category
|
334
|
-
having a high probability, the safety rating returned would have
|
335
|
-
category equal to `HARASSMENT` and harm probability set to `HIGH`.
|
336
|
-
|
337
|
-
[role="mt-4"]
|
338
|
-
=== Safety settings in Google AI Studio
|
339
|
-
|
340
|
-
You can also adjust safety settings in Google AI Studio, but you cannot
|
341
|
-
turn them off. To do so, in the *Run settings*, click *Edit safety
|
342
|
-
settings*:
|
343
|
-
|
344
|
-
And use the knobs to adjust each setting:
|
345
|
-
|
346
|
-
|
347
|
-
A [.material-icons]#warning# *No Content* message appears if the content
|
348
|
-
is blocked. To see more details, hold the pointer over *No Content* and
|
349
|
-
click [.material-icons]#warning# *Safety*.
|
350
|
-
|
351
|
-
[role="mt-4"]
|
352
|
-
=== Code examples
|
353
|
-
|
354
|
-
This section shows how to use the safety settings in code using the
|
355
|
-
python client library.
|
356
|
-
|
357
|
-
[role="mt-4"]
|
358
|
-
==== Request example
|
359
|
-
|
360
|
-
The following is a python code snippet showing how to set safety
|
361
|
-
settings in your `GenerateContent` call. This sets the harm categories
|
362
|
-
`Harassment` and `Hate speech` to `BLOCK_LOW_AND_ABOVE` which blocks any
|
363
|
-
content that has a low or higher probability of being harassment or hate
|
364
|
-
speech.
|
365
|
-
|
366
|
-
[source, js]
|
367
|
-
----
|
368
|
-
import { GoogleGenerativeAI } from "//esm.run/@google/generative-ai";
|
369
|
-
|
370
|
-
import { HarmCategory, HarmBlockThreshold } from "//esm.run/@google/generative-ai";
|
371
|
-
const safetySettings = [
|
372
|
-
{
|
373
|
-
category: HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT,
|
374
|
-
threshold: HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE
|
375
|
-
},
|
376
|
-
{
|
377
|
-
category: HarmCategory.HARM_CATEGORY_HARASSMENT,
|
378
|
-
threshold: HarmBlockThreshold.BLOCK_ONLY_HIGH
|
379
|
-
},
|
380
|
-
{
|
381
|
-
category: HarmCategory.HARM_CATEGORY_HATE_SPEECH,
|
382
|
-
threshold: HarmBlockThreshold.BLOCK_ONLY_HIGH
|
383
|
-
},
|
384
|
-
{
|
385
|
-
category: HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT,
|
386
|
-
threshold: HarmBlockThreshold.BLOCK_NONE
|
387
|
-
}
|
388
|
-
];
|
389
|
-
|
390
|
-
const model = genAI.getGenerativeModel({
|
391
|
-
model: "gemini-pro",
|
392
|
-
safetySettings
|
393
|
-
});
|
394
|
-
|
395
|
-
----
|
396
|
-
|
397
|
-
[role="mt-4"]
|
398
|
-
==== Response example
|
399
|
-
|
400
|
-
The following shows a code snippet for parsing the safety feedback from
|
401
|
-
the response.
|
402
|
-
|
403
|
-
[source, python]
|
404
|
-
----
|
405
|
-
try:
|
406
|
-
print(response.text)
|
407
|
-
except ValueError:
|
408
|
-
# If the response doesn't contain text, check if the prompt was blocked.
|
409
|
-
print(response.prompt_feedback)
|
410
|
-
# Also check the finish reason to see if the response was blocked.
|
411
|
-
print(response.candidates[0].finish_reason)
|
412
|
-
# If the finish reason was SAFETY, the safety ratings have more details.
|
413
|
-
print(response.candidates[0].safety_ratings)
|
414
|
-
----
|
415
|
-
|
416
|
-
[role="mt-4"]
|
417
|
-
=== Next steps
|
418
|
-
|
419
|
-
* See the link://ai.google.dev/api/[API reference, {browser-window--new}]
|
420
|
-
to learn more about the full API.
|
421
|
-
* Review the link://ai.google.dev/docs/safety_setting_gemini[safety guidance, {browser-window--new}]
|
422
|
-
for a general look at safety considerations when developing with LLMs.
|
423
|
-
* Learn more about assessing probability versus severity from the
|
424
|
-
link://developers.perspectiveapi.com/s/about-the-api-score[Jigsaw team, {browser-window--new}]
|
425
|
-
* Learn more about the products that contribute to safety solutions
|
426
|
-
like the
|
427
|
-
link://medium.com/jigsaw/reducing-toxicity-in-large-language-models-with-perspective-api-c31c39b7a4d7[Perspective API, {browser-window--new}].
|
428
|
-
* You can use these safety settings to create a toxicity classifier.
|
429
|
-
See the
|
430
|
-
link://ai.google.dev/examples/train_text_classifier_embeddings[classification example, {browser-window--new}]
|
431
|
-
to get started.
|
432
|
-
|
433
|
-
[role="mt-5"]
|
434
|
-
== Train a Google Gemini bot using the API interface
|
435
|
-
|
436
|
-
. Set Up a Google Cloud Project
|
437
|
-
|
438
|
-
* Create a Google Cloud project if you don't already have one.
|
439
|
-
* Enable the Dialogflow API and the Gemini API.
|
440
|
-
* Create a service account with the "Dialogflow API Admin" and "Gemini API Admin" roles.
|
441
|
-
* Generate a private key for the service account.
|
442
|
-
|
443
|
-
[start=2]
|
444
|
-
. Create a Gemini Agent
|
445
|
-
|
446
|
-
* Go to the Gemini console and click "Create Agent".
|
447
|
-
* Enter a name and description for your agent.
|
448
|
-
* Select the language for your agent.
|
449
|
-
* Click "Create".
|
450
|
-
* Link Your Dialogflow Agent to Your Gemini Agent
|
451
|
-
|
452
|
-
[start=3]
|
453
|
-
. Go to the Dialogflow console and open your agent.
|
454
|
-
|
455
|
-
* Click the "Integrations" tab.
|
456
|
-
* Click "Add Integration".
|
457
|
-
* Select "Gemini" from the list of integrations.
|
458
|
-
* Enter the Gemini API endpoint and your service account's private key.
|
459
|
-
* Click "Link".
|
460
|
-
|
461
|
-
[start=4]
|
462
|
-
. Train Your Gemini Bot
|
463
|
-
|
464
|
-
* Open the Gemini console and go to the "Training" tab.
|
465
|
-
* Click "Create Training Run".
|
466
|
-
* Select your agent from the dropdown list.
|
467
|
-
* Enter a name for the training run.
|
468
|
-
* Select the training data to use.
|
469
|
-
* Click "Start Training".
|
470
|
-
|
471
|
-
[start=5]
|
472
|
-
. Test the Agent
|
473
|
-
|
474
|
-
* Click "Test" in the top right corner of the Dialogflow Console.
|
475
|
-
* Type in a message and press "Enter."
|
476
|
-
* The agent will generate a response based on the training data.
|
477
|
-
|
478
|
-
[start=6]
|
479
|
-
. Deploy the Agent to Gemini
|
480
|
-
|
481
|
-
* Click the "Deploy" button in the top right corner of the Dialogflow Console.
|
482
|
-
* Select "Gemini" as the deployment target.
|
483
|
-
* Click "Deploy."
|
484
|
-
|
485
|
-
[role="mt-4"]
|
486
|
-
=== Monitor the Training Process
|
487
|
-
|
488
|
-
The training process will take some time. You can monitor the progress in the Gemini console.
|
489
|
-
Once the training is complete, you will receive an email notification.
|
490
|
-
Deploy Your Gemini Bot
|
491
|
-
|
492
|
-
Once the training is complete, you can deploy your bot to your website or mobile app.
|
493
|
-
To do this, you will need to generate a deployment token from the Gemini console.
|
494
|
-
Follow the instructions in the Gemini documentation to deploy your bot.
|
495
|
-
Test Your Bot
|
496
|
-
|
497
|
-
[role="mt-4"]
|
498
|
-
=== Test the bot
|
499
|
-
|
500
|
-
Once your bot is deployed, you can test it by sending it messages.
|
501
|
-
You can use the Gemini console or the Gemini API to send messages to your bot.
|
502
|
-
Evaluate the bot's responses to ensure that it is functioning as expected.
|
503
|
-
Maintain and Update Your Bot
|
504
|
-
|
505
|
-
Over time, you may need to maintain and update your bot.
|
506
|
-
This may involve retraining the bot with new data or fixing bugs.
|
507
|
-
You can use the Gemini console or the Gemini API to manage your bot.
|
508
|
-
By following these steps, you can train, deploy, and maintain a Google Gemini bot using the API interface.
|
509
|
-
|
510
|
-
[role="mt-5"]
|
511
|
-
== What's next
|
512
|
-
|
513
|
-
Prompt design is the process of creating prompts that elicit the desired
|
514
|
-
response from language models. Writing well structured prompts is an essential
|
515
|
-
part of ensuring accurate, high quality responses from a language model.
|
516
|
-
Learn about link://ai.google.dev/docs/prompt_best_practices[best practices for prompt writing, {browser-window--new}].
|
517
|
-
|
518
|
-
Gemini offers several model variations to meet the needs of different use
|
519
|
-
cases, such as input types and complexity, implementations for chat or other
|
520
|
-
dialog language tasks, and size constraints. Learn about the
|
521
|
-
link://ai.google.dev/models/gemini[available Gemini models, {browser-window--new}].
|
522
|
-
|
523
|
-
Gemini offers options for
|
524
|
-
link://ai.google.dev/docs/increase_quota[requesting rate limit increases, {browser-window--new}].
|
525
|
-
The rate limit for Gemini Pro models is 60 requests per minute (RPM).
|