noznarb commited on
Commit
284253a
1 Parent(s): 07baa2c

Upload 6 files

Browse files
.gitattributes CHANGED
@@ -70,3 +70,5 @@ ShareGPT_Vicuna_unfiltered/Experimental/removals.log filter=lfs diff=lfs merge=l
70
  ShareGPT_Vicuna_unfiltered/Experimental/ShareGPT_2023.05.04v_NoUnicode_Edition.json filter=lfs diff=lfs merge=lfs -text
71
  ShareGPT_Vicuna_unfiltered/Experimental/ShareGPT_2023.05.04v1_Nano_Wasteland.json filter=lfs diff=lfs merge=lfs -text
72
  ShareGPT_Vicuna_unfiltered/ShareGPT_2023.05.08v0_Wasteland_Edition.json filter=lfs diff=lfs merge=lfs -text
 
 
 
70
  ShareGPT_Vicuna_unfiltered/Experimental/ShareGPT_2023.05.04v_NoUnicode_Edition.json filter=lfs diff=lfs merge=lfs -text
71
  ShareGPT_Vicuna_unfiltered/Experimental/ShareGPT_2023.05.04v1_Nano_Wasteland.json filter=lfs diff=lfs merge=lfs -text
72
  ShareGPT_Vicuna_unfiltered/ShareGPT_2023.05.08v0_Wasteland_Edition.json filter=lfs diff=lfs merge=lfs -text
73
+ bluemoon_roleplay_300k_vicuna/bluemoon_roleplay_300k_vicuna.json filter=lfs diff=lfs merge=lfs -text
74
+ bluemoon_roleplay_300k_vicuna/RawData/raw.json filter=lfs diff=lfs merge=lfs -text
bluemoon_roleplay_300k_vicuna/RawData/raw.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d2ac077369e37f394b2158c9a100371c729811e7966a18c068cb04e7c2e45159
3
+ size 301734369
bluemoon_roleplay_300k_vicuna/bluemoon_roleplay_300k_vicuna.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6c734071377d534373e0b2abe859de6c21e85116baa58f2bdb36fc52c3308b1
3
+ size 241049697
bluemoon_roleplay_300k_vicuna/clean_bluemoon.py ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import json
2
+ import re
3
+
4
+ fname = "bluemoon_roleplay_300k_vicuna.json";
5
+ with open(fname,"r") as f:
6
+ d = json.load(f);
7
+
8
+ mr = re.compile(r'(((OOC|OCC|ooc|occ|Occ|Ooc|OoC))|(OFF)):.*?((([ \.!?](IC|DE|BIC|ic|bic)[:\. ])|((BiC[:]?)|[\.!?](on:|ON:|On:)))|$)');
9
+
10
+ #remaining implicit OOC'ing
11
+ manual = [
12
+ "OOC: Sorry for the redonculous wait. Work takes priority ",
13
+ "OOC: I hope you don't mind that I put a time skip in here",
14
+ "OOC: Give her broken english.",
15
+ "OoC: aww, and here I'd been hoping that something would be done with the soul edge fragments",
16
+ "OOC:Oh sure. It's ok ",
17
+ "OOC: Sorry for being gone so long...I hope you will continue rping with me",
18
+ ];
19
+
20
+ for conv in d:
21
+ for m in conv["conversations"]:
22
+ l = mr.search(m["value"]);
23
+ if l != None:
24
+ print(m["value"]);
25
+ s = mr.sub("",m["value"]);
26
+ if len(s) > 0:
27
+ print("--->",s,end="\n\n");
28
+ m["value"] = s;
29
+ else:
30
+ print("---> MANUAL INTERVENTION REQUIRED. CHECK KEYWORDS.");
31
+
32
+ s = m["value"];
33
+ for e in manual:
34
+ s = re.sub(e,"",s);
35
+ m["value"] = s;
36
+
37
+ fname = "bluemoon_roleplay_300k_vicuna_filtered.json";
38
+ with open(fname,"w") as f:
39
+ json.dump(d,f,indent=2);
bluemoon_roleplay_300k_vicuna/convert-to-vicuna.js ADDED
@@ -0,0 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ const fs = require("fs");
2
+ const data = JSON.parse(fs.readFileSync('two-chatter-only.json', 'utf8'));
3
+ let destArray = [];
4
+
5
+ // Group the contents by thread_href
6
+ const groupBy = (array, key) =>
7
+ {
8
+ return array.reduce((result, item) =>
9
+ {
10
+ (result[item[key]] = result[item[key]] || []).push(item);
11
+ return result;
12
+ }, {});
13
+ };
14
+
15
+ const grouped = groupBy(data, 'thread_href');
16
+
17
+
18
+ // Loop through each group and change the message_username field
19
+ for (let thread in grouped)
20
+ {
21
+ let messages = grouped[thread];
22
+ let destObj = {};
23
+ destObj.id = generateId();
24
+ destObj.conversations = [];
25
+
26
+ for (let i = 0; i < messages.length; i++)
27
+ {
28
+ let message = messages[i];
29
+ let conversation = {};
30
+ //console.log("Adding: " + message.message);
31
+ if (i % 2 == 0)
32
+ {
33
+ conversation.from = "human";
34
+ }
35
+ else
36
+ {
37
+ conversation.from = "gpt";
38
+ }
39
+ conversation.value = message.message;
40
+
41
+ destObj.conversations.push(conversation);
42
+ }
43
+
44
+ destArray.push(destObj);
45
+ }
46
+
47
+
48
+ fs.writeFileSync("vicuna-output.json", JSON.stringify(destArray, null, 2), "utf8");
49
+
50
+ // This is just a best guess at ID formats based on the ShareGPT format
51
+ function generateId()
52
+ {
53
+ let chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
54
+
55
+ let id = "";
56
+ for (let i = 0; i < 6; i++)
57
+ {
58
+ // Get a random index from the possible characters
59
+ let index = Math.floor(Math.random() * 62);
60
+ // Append the character at that index to the id string
61
+ id += chars[index];
62
+ }
63
+ //Append a random _XX to the id.
64
+ id += "_" + Math.floor(Math.random() * 100);
65
+
66
+ return id;
67
+ }
bluemoon_roleplay_300k_vicuna/merge_json.js ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ const fs = require('fs');
2
+
3
+ function loadAndConcatJSON(files, callback)
4
+ {
5
+ let data = [];
6
+ for (let file of files)
7
+ {
8
+ data.push(require(file));
9
+ }
10
+
11
+ let output = [].concat(...data);
12
+ output = JSON.stringify(output);
13
+
14
+ fs.writeFile('output.json', output, err =>
15
+ {
16
+ if (err)
17
+ {
18
+ callback(err);
19
+ } else
20
+ {
21
+ callback(null, output);
22
+ }
23
+ });
24
+ }
25
+
26
+
27
+ loadAndConcatJSON(['./filtered-vicuna-formatted.json', './filtered.json'], (err, result) => {
28
+ if (err)
29
+ {
30
+ console.error(err);
31
+ }
32
+ else
33
+ {
34
+ console.log("Merged JSON files.");
35
+ }
36
+ });
bluemoon_roleplay_300k_vicuna/parsejson.js ADDED
@@ -0,0 +1,196 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ // Load the fs module to read files
2
+ const fs = require('fs');
3
+ // Define the path to the json file
4
+ const filePath = 'bluemoon_roleplay_300k_vicuna.json';
5
+ const csvPath = 'data/rawData.csv';
6
+
7
+ cleanLineEndings();
8
+ validate();
9
+
10
+ function cleanLineEndings()
11
+ {
12
+ fs.readFile(filePath, 'utf8', (err, data) =>
13
+ {
14
+ const json = JSON.parse(data);
15
+ for (const entry of json)
16
+ {
17
+ for (const conversation of entry["conversations"])
18
+ {
19
+ conversation["value"] = conversation["value"].replace("\"\r", "");
20
+ }
21
+ }
22
+
23
+ const output = JSON.stringify(json, null, 2);
24
+ fs.writeFile('./cleanedup.json', output, 'utf8', (err) =>
25
+ {
26
+ });
27
+ });
28
+ }
29
+
30
+ function convertToJson()
31
+ {
32
+ fs.readFile(csvPath, "utf-8", (err, data) =>
33
+ {
34
+ if (err)
35
+ {
36
+ console.error(err);
37
+ return;
38
+ }
39
+ // Use the csvToJson function to convert the data to json
40
+ const json = csvToJson(data);
41
+ fs.writeFile("data/rawData.json", json, (err) =>
42
+ {
43
+ if (err)
44
+ {
45
+ console.error(err);
46
+ return;
47
+ }
48
+ console.log("File saved successfully");
49
+ });
50
+
51
+ // Do something with the json
52
+ });
53
+ }
54
+
55
+ //Read the file asynchronously and parse it as json
56
+ function validate()
57
+ {
58
+ fs.readFile(filePath, 'utf8', (err, data) =>
59
+ {
60
+ if (err)
61
+ {
62
+ console.error(err);
63
+ return;
64
+ }
65
+ const json = JSON.parse(data);
66
+
67
+ // Create an empty object to store the results
68
+ const result = {};
69
+
70
+ // Loop through each entry in the json array
71
+ for (const entry of json)
72
+ {
73
+ // Get the url and message_username fields from the entry
74
+ const url = entry.thread_href;
75
+ const message_username = entry.message_username;
76
+
77
+ // If the url is not already in the result object, create an empty set for it
78
+ if (!result[url])
79
+ {
80
+ result[url] = new Set();
81
+ }
82
+
83
+ // Add the message_username to the set for the url
84
+ result[url].add(message_username);
85
+ }
86
+
87
+ // Loop through each url in the result object and get the size of the set
88
+ let foundThreads = 0;
89
+ for (const url in result)
90
+ {
91
+ // Get the set of message_usernames for the url
92
+ const usernames = result[url];
93
+
94
+ // Get the count of unique message_usernames by getting the size of the set
95
+ const count = usernames.size;
96
+
97
+ // Print the url and the count
98
+ if (count > 2)
99
+ {
100
+ foundThreads++;
101
+ console.log(`${url}: ${count} `);
102
+ }
103
+ }
104
+ console.log("Threads non-two chatters: " + foundThreads);
105
+ });
106
+ }
107
+
108
+ function removedNonTwoConvos()
109
+ {
110
+ let removedCount = 0;
111
+ fs.readFile(filePath, 'utf8', (err, data) =>
112
+ {
113
+ if (err)
114
+ {
115
+ console.error(err);
116
+ return;
117
+ }
118
+ const obj = JSON.parse(data);
119
+
120
+ // Group the entries by thread_href using a Map
121
+ const map = new Map();
122
+ for (const entry of obj)
123
+ {
124
+ const key = entry.thread_href;
125
+ const value = map.get(key) || [];
126
+ value.push(entry);
127
+ map.set(key, value);
128
+ }
129
+
130
+ // Filter out the threads with more than two different message_username entries
131
+ const result = [];
132
+ for (const [key, value] of map)
133
+ {
134
+ const usernames = new Set();
135
+ for (const entry of value)
136
+ {
137
+ usernames.add(entry.message_username);
138
+ }
139
+ if (usernames.size == 2)
140
+ {
141
+ result.push(...value);
142
+ }
143
+ else
144
+ {
145
+ removedCount++;
146
+ }
147
+ }
148
+
149
+ // Stringify the result and write it to a new file
150
+ const output = JSON.stringify(result, null, 2);
151
+ fs.writeFile('./two-chatter-only.json', output, 'utf8', (err) =>
152
+ {
153
+ if (err)
154
+ {
155
+ console.error(err);
156
+ return;
157
+ }
158
+ console.log('Output written to output.json | Removed Threads: ' + removedCount);
159
+ });
160
+ });
161
+ }
162
+
163
+ // Define a function that takes a csv string as an argument
164
+ function csvToJson(csv) {
165
+ // Split the csv string by newline characters to get an array of rows
166
+ const rows = csv.split("\n");
167
+ // Get the first row as an array of headers
168
+ const headers = rows[0].replace("\r", "").split(",");
169
+
170
+ // Create an empty array to store the json objects
171
+ const json = [];
172
+
173
+ // Loop through the remaining rows
174
+ for (let i = 1; i < rows.length; i++) {
175
+ // Get the current row as an array of values
176
+ const values = rows[i].split(/,(?=(?:(?:[^"]*"){2})*[^"]*$)/);
177
+
178
+ // Create an empty object to store the key-value pairs
179
+ const obj = {};
180
+
181
+ // Loop through the headers and values and assign them to the object
182
+ for (let j = 0; j < headers.length; j++)
183
+ {
184
+ if (headers[j] !== "id" && values[j] != undefined)
185
+ {
186
+ obj[headers[j]] = values[j].replace(/^"|"$/g, "").replace(/""/g, "\"").replace("\"\r", "");
187
+ }
188
+ }
189
+
190
+ // Push the object to the json array
191
+ json.push(obj);
192
+ }
193
+
194
+ // Return the json array as a string
195
+ return JSON.stringify(json, null, 2);
196
+ }