monadic-chat 0.3.2 → 0.3.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +12 -0
- data/Gemfile.lock +8 -5
- data/README.md +44 -56
- data/apps/chat/chat.json +3 -1
- data/apps/chat/chat.md +7 -6
- data/apps/chat/chat.rb +22 -8
- data/apps/code/code.md +3 -7
- data/apps/code/code.rb +22 -8
- data/apps/linguistic/linguistic.md +2 -6
- data/apps/linguistic/linguistic.rb +22 -8
- data/apps/novel/novel.md +4 -8
- data/apps/novel/novel.rb +22 -8
- data/apps/translate/translate.md +5 -9
- data/apps/translate/translate.rb +22 -8
- data/assets/gpt2.bin +0 -0
- data/bin/monadic-chat +33 -3
- data/doc/img/how-research-mode-works.svg +1 -1
- data/doc/img/research-mode-template.svg +1 -1
- data/lib/monadic_app.rb +10 -3
- data/lib/monadic_chat/formatting.rb +9 -1
- data/lib/monadic_chat/internals.rb +58 -13
- data/lib/monadic_chat/menu.rb +2 -1
- data/lib/monadic_chat/open_ai.rb +3 -14
- data/lib/monadic_chat/tools.rb +63 -0
- data/lib/monadic_chat/version.rb +1 -1
- data/lib/monadic_chat.rb +89 -47
- data/monadic_chat.gemspec +2 -1
- metadata +23 -7
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: f86c68d3d48502db6f77c5081b1ce95478bc499519decf793701fbf0e42d59ac
|
|
4
|
+
data.tar.gz: 995f7f6581a2fc35710321405e29f1bdcbe3b3b4e00a5e2d6a543539d21d3a23
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: cbbe3b7be1cfbbf2d7144fb1e1bf92602f8a0a1d573541b4e2832c2bc7badfe9a0cd6d35efe390c41bdc6f43d3d65cd765680a1455263215c1b9d24ac5d70aa4
|
|
7
|
+
data.tar.gz: 0eed2c83c5b67545942b20a07e7096e4328b71a6acbd6f573a07f2013996dac786cce043d97cab34a6b376c284aaed59a269c45d0988e6bb0e2fe3e9528407c9
|
data/CHANGELOG.md
CHANGED
|
@@ -17,3 +17,15 @@
|
|
|
17
17
|
## [0.2.1] - 2023-03-21
|
|
18
18
|
|
|
19
19
|
- GPT-4 models supported (in `normal` mode)
|
|
20
|
+
|
|
21
|
+
## [0.3.0] - 2023-03-24
|
|
22
|
+
|
|
23
|
+
- `Research` mode now supports chat API in addition to text-completion API
|
|
24
|
+
|
|
25
|
+
## [0.3.3] - 2023-03-26
|
|
26
|
+
|
|
27
|
+
- Command line options to directly run individual apps
|
|
28
|
+
|
|
29
|
+
## [0.3.4] - 2023-03-30
|
|
30
|
+
|
|
31
|
+
- `Chat` app now supports web searches and allows users to talk about recent events
|
data/Gemfile.lock
CHANGED
|
@@ -1,12 +1,12 @@
|
|
|
1
1
|
PATH
|
|
2
2
|
remote: .
|
|
3
3
|
specs:
|
|
4
|
-
monadic-chat (0.3.
|
|
4
|
+
monadic-chat (0.3.4)
|
|
5
|
+
blingfire
|
|
5
6
|
http
|
|
6
7
|
kramdown
|
|
7
8
|
launchy
|
|
8
9
|
oj
|
|
9
|
-
parallel
|
|
10
10
|
pastel
|
|
11
11
|
rouge
|
|
12
12
|
tty-box
|
|
@@ -16,12 +16,14 @@ PATH
|
|
|
16
16
|
tty-prompt
|
|
17
17
|
tty-screen
|
|
18
18
|
tty-spinner
|
|
19
|
+
wikipedia-client
|
|
19
20
|
|
|
20
21
|
GEM
|
|
21
22
|
remote: https://rubygems.org/
|
|
22
23
|
specs:
|
|
23
24
|
addressable (2.8.1)
|
|
24
25
|
public_suffix (>= 2.0.2, < 6.0)
|
|
26
|
+
blingfire (0.1.8)
|
|
25
27
|
diff-lcs (1.5.0)
|
|
26
28
|
domain_name (0.5.20190701)
|
|
27
29
|
unf (>= 0.0.5, < 1.0.0)
|
|
@@ -45,7 +47,6 @@ GEM
|
|
|
45
47
|
ffi-compiler (~> 1.0)
|
|
46
48
|
rake (~> 13.0)
|
|
47
49
|
oj (3.14.2)
|
|
48
|
-
parallel (1.22.1)
|
|
49
50
|
pastel (0.8.0)
|
|
50
51
|
tty-color (~> 0.5)
|
|
51
52
|
public_suffix (5.0.1)
|
|
@@ -61,7 +62,7 @@ GEM
|
|
|
61
62
|
rspec-expectations (3.12.2)
|
|
62
63
|
diff-lcs (>= 1.2.0, < 2.0)
|
|
63
64
|
rspec-support (~> 3.12.0)
|
|
64
|
-
rspec-mocks (3.12.
|
|
65
|
+
rspec-mocks (3.12.5)
|
|
65
66
|
diff-lcs (>= 1.2.0, < 2.0)
|
|
66
67
|
rspec-support (~> 3.12.0)
|
|
67
68
|
rspec-support (3.12.0)
|
|
@@ -103,6 +104,8 @@ GEM
|
|
|
103
104
|
unf_ext (0.0.8.2)
|
|
104
105
|
unicode-display_width (2.4.2)
|
|
105
106
|
unicode_utils (1.4.0)
|
|
107
|
+
wikipedia-client (1.17.0)
|
|
108
|
+
addressable (~> 2.7)
|
|
106
109
|
wisper (2.0.1)
|
|
107
110
|
|
|
108
111
|
PLATFORMS
|
|
@@ -116,4 +119,4 @@ DEPENDENCIES
|
|
|
116
119
|
rspec
|
|
117
120
|
|
|
118
121
|
BUNDLED WITH
|
|
119
|
-
2.4.
|
|
122
|
+
2.4.9
|
data/README.md
CHANGED
|
@@ -10,59 +10,18 @@
|
|
|
10
10
|
<kbd><img src="https://user-images.githubusercontent.com/18207/225505520-53e6f2c4-84a8-4128-a005-3fe980ec2449.gif" width="100%" /></kbd>
|
|
11
11
|
</p>
|
|
12
12
|
|
|
13
|
-
> **
|
|
14
|
-
> This software is
|
|
13
|
+
> **Note**
|
|
14
|
+
> This software is *work in progress* and *under active development*. It may be unstable, and the latest version may behave slightly differently than this document. Also, specifications may change in the future.
|
|
15
15
|
|
|
16
16
|
**Change Log**
|
|
17
17
|
|
|
18
|
-
- [March
|
|
18
|
+
- [March 30, 2023] `Chat` app now supports web searches and allows users to talk about recent events
|
|
19
|
+
- [March 26, 2023] Command line options to directly run individual apps
|
|
20
|
+
- [March 24, 2023] `Research` mode now supports chat API in addition to text-completion API
|
|
19
21
|
- [March 21, 2023] GPT-4 models supported (in `normal` mode)
|
|
20
22
|
- [March 20, 2023] Text and figure in "How the research mode workds" section updated
|
|
21
23
|
- [March 13, 2023] Text on the architecture of the `research` mode updated in accordance with Version 0.2.0
|
|
22
24
|
|
|
23
|
-
## Table of Contents
|
|
24
|
-
|
|
25
|
-
## TOC
|
|
26
|
-
|
|
27
|
-
- [Table of Contents](#table-of-contents)
|
|
28
|
-
- [TOC](#toc)
|
|
29
|
-
- [Introduction](#introduction)
|
|
30
|
-
- [Dependencies](#dependencies)
|
|
31
|
-
- [Installation](#installation)
|
|
32
|
-
- [Using RubyGems](#using-rubygems)
|
|
33
|
-
- [Clone the GitHub Repository](#clone-the-github-repository)
|
|
34
|
-
- [Usage](#usage)
|
|
35
|
-
- [Authentication](#authentication)
|
|
36
|
-
- [Select Main Menu Item](#select-main-menu-item)
|
|
37
|
-
- [Roles](#roles)
|
|
38
|
-
- [System-Wide Functions](#system-wide-functions)
|
|
39
|
-
- [Apps](#apps)
|
|
40
|
-
- [Chat](#chat)
|
|
41
|
-
- [Code](#code)
|
|
42
|
-
- [Novel](#novel)
|
|
43
|
-
- [Translate](#translate)
|
|
44
|
-
- [Modes](#modes)
|
|
45
|
-
- [Normal Mode](#normal-mode)
|
|
46
|
-
- [Research Mode](#research-mode)
|
|
47
|
-
- [What is Research Mode?](#what-is-research-mode)
|
|
48
|
-
- [How Research Mode Works](#how-research-mode-works)
|
|
49
|
-
- [Accumulator](#accumulator)
|
|
50
|
-
- [Reducer](#reducer)
|
|
51
|
-
- [Creating New App](#creating-new-app)
|
|
52
|
-
- [File Structure](#file-structure)
|
|
53
|
-
- [Reducer Code](#reducer-code)
|
|
54
|
-
- [Monadic Chat Template](#monadic-chat-template)
|
|
55
|
-
- [Extra Template for `Research` Mode](#extra-template-for-research-mode)
|
|
56
|
-
- [What is Monadic about Monadic Chat?](#what-is-monadic-about-monadic-chat)
|
|
57
|
-
- [Unit, Map, and Join](#unit-map-and-join)
|
|
58
|
-
- [Discourse Management Object](#discourse-management-object)
|
|
59
|
-
- [Future Plans](#future-plans)
|
|
60
|
-
- [Bibliographical Data](#bibliographical-data)
|
|
61
|
-
- [Acknowledgments](#acknowledgments)
|
|
62
|
-
- [Contributing](#contributing)
|
|
63
|
-
- [Author](#author)
|
|
64
|
-
- [License](#license)
|
|
65
|
-
|
|
66
25
|
## Introduction
|
|
67
26
|
|
|
68
27
|
**Monadic Chat** is a user-friendly command-line client application that utilizes OpenAI’s Text Completion API and Chat API to facilitate ChatGPT-style conversations with OpenAI’s large language models (LLM) on any terminal application of your choice.
|
|
@@ -71,15 +30,25 @@ The conversation history can be saved in a JSON file, which can be loaded later
|
|
|
71
30
|
|
|
72
31
|
Monadic Chat includes four pre-built apps (`Chat`, `Code`, `Novel`, and `Translate`) that are designed for different types of discourse through interactive conversation with the LLM. Users also have the option to create their own apps by writing new templates.
|
|
73
32
|
|
|
33
|
+
Monadic Chat's `normal` mode enables ChatGPT-like conversations on the command line. The `research` mode has a mechanism to handle various related information as "state" behind the conversation. This allows, for example, to retrieve the current conversation *topic* at each utterance turn, and to keep its development as a list.
|
|
34
|
+
|
|
74
35
|
## Dependencies
|
|
75
36
|
|
|
76
37
|
- Ruby 2.6.10 or greater
|
|
77
38
|
- OpenAI API Token
|
|
78
39
|
- A command line terminal app such as:
|
|
79
40
|
- Terminal or [iTerm2](https://iterm2.com/) (MacOS)
|
|
80
|
-
- [Windows Terminal](https://apps.microsoft.com/store/detail/windows-terminal) (Windows 11)
|
|
81
|
-
- GNOME Terminal (Linux)
|
|
82
41
|
- [Alacritty](https://alacritty.org/) (Multi-platform)
|
|
42
|
+
- [Windows Terminal](https://apps.microsoft.com/store/detail/windows-terminal) (Windows)
|
|
43
|
+
- GNOME Terminal (Linux)
|
|
44
|
+
|
|
45
|
+
> **Note on Using Monadic Chat on Windows**
|
|
46
|
+
> Monadic Chat does not support running on Windows, but you can install and use Linux Destribution on WSL2. Or you can use it without WSL2 by following these steps:
|
|
47
|
+
>
|
|
48
|
+
> 1. install Windows Terminal
|
|
49
|
+
> 2. install [Git Bash](https://gitforwindows.org/) (make sure to check the `Install profile for Windows Terminal` checkbox
|
|
50
|
+
> 3. install Ruby with [Ruby Installer](https://rubyinstaller.org/)
|
|
51
|
+
> 4. Open Windows Terminal with Git Bash profile and follow the instruction below.
|
|
83
52
|
|
|
84
53
|
## Installation
|
|
85
54
|
|
|
@@ -147,7 +116,7 @@ Once the correct access token is verified, the access token is saved in the conf
|
|
|
147
116
|
|
|
148
117
|
`$HOME/monadic_chat.conf`
|
|
149
118
|
|
|
150
|
-
###
|
|
119
|
+
### Main Menu
|
|
151
120
|
|
|
152
121
|
Upon successful authentication, a menu to select a specific app will appear. Each app generates different types of text through an interactive chat-style conversation between the user and the AI. Four apps are available by default: [`chat`](#chat), [`code`](#code), [`novel`](#novel), and [`translate`](#translate).
|
|
153
122
|
|
|
@@ -163,6 +132,29 @@ Selecting `readme` will take you to the README on the GitHub repository (the doc
|
|
|
163
132
|
|
|
164
133
|
In the main menu, you can use the cursor keys and the enter key to make a selection. You can also narrow down the choices each time you type a letter.
|
|
165
134
|
|
|
135
|
+
### Direct Commands
|
|
136
|
+
|
|
137
|
+
The following commands can be entered to start each app directly on the command line, without using the main menu.
|
|
138
|
+
|
|
139
|
+
```
|
|
140
|
+
monadic-chat <app-name>
|
|
141
|
+
```
|
|
142
|
+
|
|
143
|
+
Each of the four standard applications can be launched as follows. When launched, an interactive chat interface appears.
|
|
144
|
+
|
|
145
|
+
```
|
|
146
|
+
monadic-chat chat
|
|
147
|
+
monadic-chat code
|
|
148
|
+
monadic-chat novel
|
|
149
|
+
monadic-chat translate
|
|
150
|
+
```
|
|
151
|
+
|
|
152
|
+
You can also give text input directly to each app in the following format and get only a response to it (without starting the interactive chat interface)
|
|
153
|
+
|
|
154
|
+
```
|
|
155
|
+
monadic-chat <app-name> <input-text>
|
|
156
|
+
```
|
|
157
|
+
|
|
166
158
|
### Roles
|
|
167
159
|
|
|
168
160
|
Each message in the conversation is labeled with one of three roles: `User`, `GPT`, or `System`.
|
|
@@ -198,7 +190,7 @@ For detailed information on each parameter, please refer to OpenAI's [API Docume
|
|
|
198
190
|
|
|
199
191
|
**data/context**
|
|
200
192
|
|
|
201
|
-
In `normal` mode, this function only displays the conversation history between User and GPT. In `research` mode, metadata (e.g., topics, language being used, number of turns) values are presented.
|
|
193
|
+
In `normal` mode, this function only displays the conversation history between User and GPT. In `research` mode, metadata (e.g., topics, language being used, number of turns) values are presented. In addition to the metadata returned in the API response, the approximate number of tokens in the current template is also displayed.
|
|
202
194
|
|
|
203
195
|
Program code in the conversation history will be syntax highlighted (if possible). The same applies to output via the `html` command available from the function menu.
|
|
204
196
|
|
|
@@ -303,7 +295,7 @@ In the default configuration, the dialogue messages are reduced after ten turns
|
|
|
303
295
|
|
|
304
296
|
The current default language model for `research` mode is `gpt-3.5-turbo`.
|
|
305
297
|
|
|
306
|
-
In `research` mode, the conversation between the user and the large-scale language model is accomplished
|
|
298
|
+
In `research` mode, the conversation between the user and the large-scale language model is accomplished with a mechanism that tracks the conversation history in a monadic structure. In the default configuration, the dialogue messages are reduced after ten turns by deleting the oldest ones (but not the messages that the `system` role has given as instructions).
|
|
307
299
|
|
|
308
300
|
If you wish to specify how the conversation history is handled as the interaction with the GPT model unfolds, you can write a `Proc` object containing Ruby code. Since various metadata are available in this mode, finer-grained control is possible.
|
|
309
301
|
|
|
@@ -403,8 +395,6 @@ Below is a sample HTML displaying the conversation (paris of an input sentence a
|
|
|
403
395
|
|
|
404
396
|
<br />
|
|
405
397
|
|
|
406
|
-
|
|
407
|
-
|
|
408
398
|
### File Structure
|
|
409
399
|
|
|
410
400
|
New Monadic Chat apps must be placed inside the `apps` folder. The folders and files for default apps `chat`, `code`, `novel`, and `translate` are also in this folder.
|
|
@@ -514,7 +504,6 @@ Monadic Chat replaces `{{MESSAGES}}` with messages from past conversations when
|
|
|
514
504
|
"prompt": "\"We didn't have a camera.\"",
|
|
515
505
|
"response": "`[S [NP We] [VP [V didn't] [VP [V have] [NP [Det a] [N camera] ] ] ] ] ]`\n\n###\n\n",
|
|
516
506
|
"mode": "linguistic",
|
|
517
|
-
"tokens": 351
|
|
518
507
|
"turns": 3,
|
|
519
508
|
"sentence_type": ["declarative"],
|
|
520
509
|
"sentiment": ["sad"],
|
|
@@ -526,7 +515,7 @@ This is the core of the extra template for `research` mode.
|
|
|
526
515
|
|
|
527
516
|
Note that the extra template is written in Markdown format, so the above JSON object is actually separated from the rest of the template as a [fenced code block](https://www.markdownguide.org/extended-syntax/#fenced-code-blocks).
|
|
528
517
|
|
|
529
|
-
The required properties of this JSON object are `prompt`, `response`,
|
|
518
|
+
The required properties of this JSON object are `prompt`, `response`, and `mode`. Other properties are optional. The `mode` property is used to check the app name when saving the conversation data or loading from an external file. The `turns` property is also used in the reducer mechanism.
|
|
530
519
|
|
|
531
520
|
The JSON object in the `research` mode template is saved in the user’s home directory (`$HOME`) with the file `monadic_chat.json`. The content is overwritten every time the JSON object is updated. Note that this JSON file is created for logging purposes . Modifying its content does not affect the processes carried out by the app.
|
|
532
521
|
|
|
@@ -541,7 +530,6 @@ Make sure the following content requirements are all fulfilled:
|
|
|
541
530
|
- analyze the new prompt's sentence type and set a sentence type value such as "interrogative", "imperative", "exclamatory", or "declarative" to the "sentence_type" property
|
|
542
531
|
- analyze the new prompt's sentiment and set one or more sentiment types such as "happy", "excited", "troubled", "upset", or "sad" to the "sentiment" property
|
|
543
532
|
- summarize the user's messages so far and update the "summary" property with a text of fewer than 100 words using as many discourse markers such as "because", "therefore", "but", and "so" to show the logical connection between the events.
|
|
544
|
-
- update the value of "tokens" with the number of tokens of the resulting JSON object"
|
|
545
533
|
- increment the value of "turns" by 1
|
|
546
534
|
```
|
|
547
535
|
|
data/apps/chat/chat.json
CHANGED
|
@@ -1,8 +1,10 @@
|
|
|
1
1
|
{"messages": [
|
|
2
2
|
{"role": "system",
|
|
3
|
-
"content": "You are a friendly but professional consultant
|
|
3
|
+
"content": "You are a friendly but professional consultant having real-time, up-to-date, information about almost anything. You are able to answer various types of questions, writes computer program code, makes decent suggestions, and gives helpful advice in response to a prompt from the user.\n\nThe date today is {{DATE}}.\n\nIf the prompt is not clear enough, ask the user to rephrase it. You are able to empathize with the user; insert an emoji (displayable on the terminal screen) that you deem appropriate for the user's input at the beginning of your response."},
|
|
4
4
|
{"role": "user",
|
|
5
5
|
"content": "Can I ask something?"},
|
|
6
6
|
{"role": "assistant",
|
|
7
7
|
"content": "Sure!"}
|
|
8
8
|
]}
|
|
9
|
+
|
|
10
|
+
|
data/apps/chat/chat.md
CHANGED
|
@@ -1,6 +1,7 @@
|
|
|
1
1
|
{{SYSTEM}}
|
|
2
2
|
|
|
3
3
|
Create a response to "NEW PROMPT" from the user and set your response to the "response" property of the JSON object below. The preceding conversation is stored in "PAST MESSAGES".
|
|
4
|
+
|
|
4
5
|
The preceding conversation is stored in "PAST MESSAGES". In "PAST MESSAGES", "assistant" refers to you. Make your response as detailed as possible.
|
|
5
6
|
|
|
6
7
|
NEW PROMPT: {{PROMPT}}
|
|
@@ -12,33 +13,33 @@ JSON:
|
|
|
12
13
|
|
|
13
14
|
```json
|
|
14
15
|
{
|
|
15
|
-
"prompt": "Can I ask something?",
|
|
16
16
|
"response": "Sure!",
|
|
17
|
+
"summary": "",
|
|
17
18
|
"mode": "chat",
|
|
18
|
-
"turns": 1,
|
|
19
19
|
"language": "English",
|
|
20
20
|
"topics": [],
|
|
21
|
-
"
|
|
21
|
+
"confidence": 1.00,
|
|
22
|
+
"ambiguity": 0.00
|
|
22
23
|
}
|
|
23
24
|
```
|
|
24
25
|
|
|
25
26
|
Make sure the following content requirements are all fulfilled:
|
|
26
27
|
|
|
27
28
|
- keep the value of the "mode" property at "chat"
|
|
28
|
-
- set the new prompt to the "prompt" property
|
|
29
29
|
- create your response to the new prompt based on the PAST MESSAGES and set it to "response"
|
|
30
30
|
- if the new prompt is in a language other than the current value of "language", set the name of the new prompt language to "language" and make sure that "response" is in that language
|
|
31
31
|
- make your response in the same language as the new prompt
|
|
32
32
|
- analyze the topic of the new prompt and insert it at the end of the value list of the "topics" property
|
|
33
|
+
- summarize the user's messages so far and update the "summary" property with a text of fewer than 100 words
|
|
34
|
+
- update the value of the "confidence" property based on the factuality of your response, ranging from 0.00 (not at all confident) to 1.00 (fully confident)
|
|
35
|
+
- update the value of the "ambiguity" property based on the clarity of the user input, ranging from 0.00 (not at all ambiguous, clearly stated) to 1.00 (fully ambiguous, nonsensical)
|
|
33
36
|
- avoid giving a response that is the same or similar to one of the previous responses in PAST MESSAGES
|
|
34
37
|
- program code in the response must be embedded in a code block in the markdown text
|
|
35
|
-
- update the value of "tokens" with the number of tokens of the resulting JSON object"
|
|
36
38
|
|
|
37
39
|
Make sure the following formal requirements are all fulfilled:
|
|
38
40
|
|
|
39
41
|
- do not use invalid characters in the JSON object
|
|
40
42
|
- escape double quotes and other special characters in the text values in the resulting JSON object
|
|
41
|
-
- increment the value of "turns" by 1
|
|
42
43
|
- check the validity of the generated JSON object and correct any possible parsing problems before returning it
|
|
43
44
|
|
|
44
45
|
Return your response consisting solely of the JSON object wrapped in "<JSON>\n" and "\n</JSON>" tags.
|
data/apps/chat/chat.rb
CHANGED
|
@@ -15,8 +15,8 @@ class Chat < MonadicApp
|
|
|
15
15
|
"top_p" => 1.0,
|
|
16
16
|
"presence_penalty" => 0.2,
|
|
17
17
|
"frequency_penalty" => 0.2,
|
|
18
|
-
"model" =>
|
|
19
|
-
"max_tokens" =>
|
|
18
|
+
"model" => research_mode ? SETTINGS["research_model"] : SETTINGS["normal_model"],
|
|
19
|
+
"max_tokens" => 1000,
|
|
20
20
|
"stream" => stream,
|
|
21
21
|
"stop" => nil
|
|
22
22
|
}.merge(params)
|
|
@@ -38,14 +38,23 @@ class Chat < MonadicApp
|
|
|
38
38
|
# @messages: messages to this point #
|
|
39
39
|
# @metadata: currently available metdata sent from GPT #
|
|
40
40
|
############################################################
|
|
41
|
-
|
|
42
41
|
conditions = [
|
|
43
42
|
@messages.size > 1,
|
|
44
|
-
@
|
|
43
|
+
@messages.size > @num_retained_turns * 2 + 1
|
|
45
44
|
]
|
|
46
45
|
|
|
47
|
-
|
|
48
|
-
|
|
46
|
+
if conditions.all?
|
|
47
|
+
to_delete = []
|
|
48
|
+
new_num_messages = @messages.size
|
|
49
|
+
@messages.each_with_index do |ele, i|
|
|
50
|
+
if ele["role"] != "system"
|
|
51
|
+
to_delete << i
|
|
52
|
+
new_num_messages -= 1
|
|
53
|
+
end
|
|
54
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
55
|
+
end
|
|
56
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
57
|
+
end
|
|
49
58
|
when :normal
|
|
50
59
|
############################################################
|
|
51
60
|
# Normal mode recuder defined here #
|
|
@@ -53,16 +62,21 @@ class Chat < MonadicApp
|
|
|
53
62
|
############################################################
|
|
54
63
|
|
|
55
64
|
conditions = [
|
|
65
|
+
@messages.size > 1,
|
|
56
66
|
@messages.size > @num_retained_turns * 2 + 1
|
|
57
67
|
]
|
|
58
68
|
|
|
59
69
|
if conditions.all?
|
|
70
|
+
to_delete = []
|
|
71
|
+
new_num_messages = @messages.size
|
|
60
72
|
@messages.each_with_index do |ele, i|
|
|
61
73
|
if ele["role"] != "system"
|
|
62
|
-
|
|
63
|
-
|
|
74
|
+
to_delete << i
|
|
75
|
+
new_num_messages -= 1
|
|
64
76
|
end
|
|
77
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
65
78
|
end
|
|
79
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
66
80
|
end
|
|
67
81
|
end
|
|
68
82
|
end
|
data/apps/code/code.md
CHANGED
|
@@ -11,33 +11,29 @@ JSON:
|
|
|
11
11
|
|
|
12
12
|
```json
|
|
13
13
|
{
|
|
14
|
-
"prompt": "Can I ask something?",
|
|
15
14
|
"response": "Sure!",
|
|
15
|
+
"summary": "",
|
|
16
16
|
"mode": "chat",
|
|
17
|
-
"turns": 1,
|
|
18
17
|
"language": "English",
|
|
19
|
-
"topics": []
|
|
20
|
-
"tokens": 109
|
|
18
|
+
"topics": []
|
|
21
19
|
}
|
|
22
20
|
```
|
|
23
21
|
|
|
24
22
|
Make sure the following content requirements are all fulfilled:
|
|
25
23
|
|
|
26
24
|
- keep the value of the "mode" property at "chat"
|
|
27
|
-
- set the new prompt to the "prompt" property
|
|
28
25
|
- create your response to the new prompt based on "PAST MESSAGES" and set it to "response"
|
|
29
26
|
- if the prompt is in a language other than the current value of "language", set the name of the new prompt language to "language" and make sure that "response" is in that language
|
|
30
27
|
- make your response in the same language as the new prompt
|
|
31
28
|
- analyze the topic of the new prompt and insert it at the end of the value list of the "topics" property
|
|
29
|
+
- summarize the user's messages so far and update the "summary" property with a text of fewer than 100 words
|
|
32
30
|
- avoid giving a response that is the same or similar to one of the previous responses in "PAST MESSAGES"
|
|
33
31
|
- program code in the response must be embedded in a code block in the markdown text
|
|
34
|
-
- update the value of "tokens" with the number of tokens of the resulting JSON object"
|
|
35
32
|
|
|
36
33
|
Make sure the following formal requirements are all fulfilled:
|
|
37
34
|
|
|
38
35
|
- do not use invalid characters in the JSON object
|
|
39
36
|
- escape double quotes and other special characters in the text values in the resulting JSON object
|
|
40
|
-
- increment the value of "turns" by 1
|
|
41
37
|
- check the validity of the generated JSON object and correct any possible parsing problems before returning it
|
|
42
38
|
|
|
43
39
|
Return your response consisting solely of the JSON object wrapped in "<JSON>\n" and "\n</JSON>" tags.
|
data/apps/code/code.rb
CHANGED
|
@@ -15,8 +15,8 @@ class Code < MonadicApp
|
|
|
15
15
|
"top_p" => 1.0,
|
|
16
16
|
"presence_penalty" => 0.0,
|
|
17
17
|
"frequency_penalty" => 0.0,
|
|
18
|
-
"model" =>
|
|
19
|
-
"max_tokens" =>
|
|
18
|
+
"model" => research_mode ? SETTINGS["research_model"] : SETTINGS["normal_model"],
|
|
19
|
+
"max_tokens" => 1000,
|
|
20
20
|
"stream" => stream,
|
|
21
21
|
"stop" => nil
|
|
22
22
|
}.merge(params)
|
|
@@ -38,14 +38,23 @@ class Code < MonadicApp
|
|
|
38
38
|
# @messages: messages to this point #
|
|
39
39
|
# @metadata: currently available metdata sent from GPT #
|
|
40
40
|
############################################################
|
|
41
|
-
|
|
42
41
|
conditions = [
|
|
43
42
|
@messages.size > 1,
|
|
44
|
-
@
|
|
43
|
+
@messages.size > @num_retained_turns * 2 + 1
|
|
45
44
|
]
|
|
46
45
|
|
|
47
|
-
|
|
48
|
-
|
|
46
|
+
if conditions.all?
|
|
47
|
+
to_delete = []
|
|
48
|
+
new_num_messages = @messages.size
|
|
49
|
+
@messages.each_with_index do |ele, i|
|
|
50
|
+
if ele["role"] != "system"
|
|
51
|
+
to_delete << i
|
|
52
|
+
new_num_messages -= 1
|
|
53
|
+
end
|
|
54
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
55
|
+
end
|
|
56
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
57
|
+
end
|
|
49
58
|
when :normal
|
|
50
59
|
############################################################
|
|
51
60
|
# Normal mode recuder defined here #
|
|
@@ -53,16 +62,21 @@ class Code < MonadicApp
|
|
|
53
62
|
############################################################
|
|
54
63
|
|
|
55
64
|
conditions = [
|
|
65
|
+
@messages.size > 1,
|
|
56
66
|
@messages.size > @num_retained_turns * 2 + 1
|
|
57
67
|
]
|
|
58
68
|
|
|
59
69
|
if conditions.all?
|
|
70
|
+
to_delete = []
|
|
71
|
+
new_num_messages = @messages.size
|
|
60
72
|
@messages.each_with_index do |ele, i|
|
|
61
73
|
if ele["role"] != "system"
|
|
62
|
-
|
|
63
|
-
|
|
74
|
+
to_delete << i
|
|
75
|
+
new_num_messages -= 1
|
|
64
76
|
end
|
|
77
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
65
78
|
end
|
|
79
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
66
80
|
end
|
|
67
81
|
end
|
|
68
82
|
end
|
|
@@ -13,27 +13,23 @@ JSON:
|
|
|
13
13
|
|
|
14
14
|
```json
|
|
15
15
|
{
|
|
16
|
-
"prompt": "\"We didn't have a camera.\"",
|
|
17
16
|
"response": "`[S [NP We] [VP [V didn't] [VP [V have] [NP [Det a] [N camera] ] ] ] ] ]`",
|
|
18
17
|
"mode": "linguistic",
|
|
19
|
-
"turns": 3,
|
|
20
18
|
"sentence_type": ["declarative"],
|
|
21
19
|
"sentiment": ["sad"],
|
|
22
20
|
"summary": "The user saw a beautiful sunset, but did not take a picture because the user did not have a camera.",
|
|
23
|
-
"
|
|
21
|
+
"relevance": 0.80
|
|
24
22
|
}
|
|
25
23
|
```
|
|
26
24
|
|
|
27
25
|
Make sure the following content requirements are all fulfilled:
|
|
28
26
|
|
|
29
27
|
- keep the value of the "mode" property at "linguistic"
|
|
30
|
-
- set the new prompt to the "prompt" property
|
|
31
28
|
- create your response to the new prompt based on "PAST MESSAGES" and set it to "response"
|
|
32
29
|
- analyze the new prompt's sentence type and set a sentence type value such as "interrogative", "imperative", "exclamatory", or "declarative" to the "sentence_type" property
|
|
33
30
|
- analyze the new prompt's sentiment and set one or more sentiment types such as "happy", "excited", "troubled", "upset", or "sad" to the "sentiment" property
|
|
34
31
|
- summarize the user's messages so far and update the "summary" property with a text of fewer than 100 words using as many discourse markers such as "because", "therefore", "but", and "so" to show the logical connection between the events.
|
|
35
|
-
- update the value of "
|
|
36
|
-
- increment the value of "turns" by 1
|
|
32
|
+
- update the value of the "relevance" property indicating the degree to which the new input is naturally interpreted based on previous discussions, ranging from 0.0 (extremely difficult) to 1.0 (completely easy)
|
|
37
33
|
|
|
38
34
|
Make sure the following formal requirements are all fulfilled:
|
|
39
35
|
|
|
@@ -15,8 +15,8 @@ class Linguistic < MonadicApp
|
|
|
15
15
|
"top_p" => 1.0,
|
|
16
16
|
"presence_penalty" => 0.0,
|
|
17
17
|
"frequency_penalty" => 0.0,
|
|
18
|
-
"model" =>
|
|
19
|
-
"max_tokens" =>
|
|
18
|
+
"model" => research_mode ? SETTINGS["research_model"] : SETTINGS["normal_model"],
|
|
19
|
+
"max_tokens" => 1000,
|
|
20
20
|
"stream" => stream,
|
|
21
21
|
"stop" => nil
|
|
22
22
|
}.merge(params)
|
|
@@ -38,14 +38,23 @@ class Linguistic < MonadicApp
|
|
|
38
38
|
# @messages: messages to this point #
|
|
39
39
|
# @metadata: currently available metdata sent from GPT #
|
|
40
40
|
############################################################
|
|
41
|
-
|
|
42
41
|
conditions = [
|
|
43
42
|
@messages.size > 1,
|
|
44
|
-
@
|
|
43
|
+
@messages.size > @num_retained_turns * 2 + 1
|
|
45
44
|
]
|
|
46
45
|
|
|
47
|
-
|
|
48
|
-
|
|
46
|
+
if conditions.all?
|
|
47
|
+
to_delete = []
|
|
48
|
+
new_num_messages = @messages.size
|
|
49
|
+
@messages.each_with_index do |ele, i|
|
|
50
|
+
if ele["role"] != "system"
|
|
51
|
+
to_delete << i
|
|
52
|
+
new_num_messages -= 1
|
|
53
|
+
end
|
|
54
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
55
|
+
end
|
|
56
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
57
|
+
end
|
|
49
58
|
when :normal
|
|
50
59
|
############################################################
|
|
51
60
|
# Normal mode recuder defined here #
|
|
@@ -53,16 +62,21 @@ class Linguistic < MonadicApp
|
|
|
53
62
|
############################################################
|
|
54
63
|
|
|
55
64
|
conditions = [
|
|
65
|
+
@messages.size > 1,
|
|
56
66
|
@messages.size > @num_retained_turns * 2 + 1
|
|
57
67
|
]
|
|
58
68
|
|
|
59
69
|
if conditions.all?
|
|
70
|
+
to_delete = []
|
|
71
|
+
new_num_messages = @messages.size
|
|
60
72
|
@messages.each_with_index do |ele, i|
|
|
61
73
|
if ele["role"] != "system"
|
|
62
|
-
|
|
63
|
-
|
|
74
|
+
to_delete << i
|
|
75
|
+
new_num_messages -= 1
|
|
64
76
|
end
|
|
77
|
+
break if new_num_messages <= @num_retained_turns * 2 + 1
|
|
65
78
|
end
|
|
79
|
+
@messages.delete_if.with_index { |_, i| to_delete.include? i }
|
|
66
80
|
end
|
|
67
81
|
end
|
|
68
82
|
end
|
data/apps/novel/novel.md
CHANGED
|
@@ -11,28 +11,24 @@ JSON:
|
|
|
11
11
|
|
|
12
12
|
```json
|
|
13
13
|
{
|
|
14
|
-
"prompt": "The preface to the novel is presented",
|
|
15
14
|
"response": "What follows is a story that an AI assistant tells. It is guaranteed that this will be an incredibly realistic and interesting novel.",
|
|
16
|
-
"
|
|
17
|
-
"
|
|
18
|
-
"tokens": 147
|
|
15
|
+
"summary": "",
|
|
16
|
+
"mode": "novel"
|
|
19
17
|
}
|
|
20
18
|
```
|
|
21
19
|
|
|
22
20
|
Make sure the following content requirements are all fulfilled:
|
|
23
21
|
|
|
24
22
|
- keep the value of the "mode" property at "novel"
|
|
25
|
-
- set the new prompt to the "prompt" property
|
|
26
23
|
- create your new paragraph in response to the new prompt and set it to "response"
|
|
27
24
|
- do not repeat in your response what is already told in "PAST MESSAGES"
|
|
28
|
-
-
|
|
29
|
-
-
|
|
25
|
+
- make your response as detailed as possible within the maximum limit of 200 words
|
|
26
|
+
- summarize the user's messages so far and update the "summary" property with a text of fewer than 100 words
|
|
30
27
|
|
|
31
28
|
Make sure the following formal requirements are all fulfilled:
|
|
32
29
|
|
|
33
30
|
- do not use invalid characters in the JSON object
|
|
34
31
|
- escape double quotes and other special characters in the text values in the resulting JSON object
|
|
35
|
-
- increment the value of "turns" by 1
|
|
36
32
|
- check the validity of the generated JSON object and correct any possible parsing problems before returning it
|
|
37
33
|
|
|
38
34
|
Return your response consisting solely of the JSON object wrapped in "<JSON>\n" and "\n</JSON>" tags.
|