Hugging Face's logo

Datasets:
mookiezi
/
Discord-Dialogues

Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
1M - 10M
Tags:
discord
chatml
conversation
dialogue
multi-turn
single-turn
Libraries:
Datasets
pandas
License:
Dataset card Data Studio Files
xet
Community
1
mookiezi commited on
Commit
c9c90e4
·
1 Parent(s): b2bfe3d

Remove more blanks and ToS breaking content

Browse files
Files changed (1) hide show
  1. README.md +79 -71
README.md CHANGED
@@ -27,7 +27,7 @@ size_categories:
27
 
28
  > **Discord-Dialogues** is a large-scale dataset of anonymized Discord conversations from late spring to early fall 2025 for training and evaluating realistic conversational AI models in a ChatML-friendly format.
29
 
30
- This dataset contains 7.3 million exchanges spread out over 16 million turns, with more than 139 million words.
31
 
32
  ---
33
 
@@ -111,76 +111,84 @@ The full end-to-end pipeline is documented in the [dataset-pipeline GitHub repos
111
  ## Dataset Statistics <span style="font-weight:normal;">(using the [NousResearch/Hermes-3-Llama-3.1-8B tokenizer](https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B))</span>
112
 
113
  <div style="display:flex; gap:20px; align-items:flex-start;">
114
- <div>
115
- | Metric | Value |
116
- | ---------------------- | ------------: |
117
- | Samples (count) | 7,303,464 |
118
- | Total turns | 16,881,010 |
119
- | Total assistant turns | 9,016,287 |
120
- | Min length (tokens) | 10 |
121
- | Max length (tokens) | 2,542 |
122
- | Mean length (tokens) | 32.79 |
123
- | Median length (tokens) | 28 |
124
- | Std dev (tokens) | 16.56 |
125
- | Skew | 6.04 |
126
- | Kurtosis | 326.54 |
127
- | Total tokens | 239,458,213 |
128
- | Total characters | 1,242,238,794 |
129
- | Total words | 139,922,950 |
130
- | Avg chars per sample | 170.09 |
131
- | Avg words per sample | 19.16 |
132
- | Avg chars per word | 8.88 |
133
- | Tokens per char | 0.19 |
134
- </div>
135
-
136
- <div>
137
- | Tokens | Count |
138
- | --------- | --------: |
139
- | 8–16 | 107,264 |
140
- | 16–32 | 4,278,713 |
141
- | 32–64 | 2,566,176 |
142
- | 64–128 | 334,829 |
143
- | 128–256 | 15,920 |
144
- | 256–384 | 363 |
145
- | 384–512 | 71 |
146
- | 512–768 | 78 |
147
- | 768–1024 | 30 |
148
- | 1024–2048 | 17 |
149
- | 2048–4096 | 3 |
150
- </div>
151
-
152
- <div>
153
- | Turns | Count |
154
- | ----- | --------: |
155
- | 2 | 5,795,019 |
156
- | 3 | 1,038,500 |
157
- | 4 | 304,442 |
158
- | 5 | 96,758 |
159
- | 6 | 38,620 |
160
- | 7 | 15,714 |
161
- | 8 | 7,108 |
162
- | 9 | 3,391 |
163
- | 10 | 1,709 |
164
- | 11 | 909 |
165
- | 12 | 526 |
166
- | 13 | 291 |
167
- | 14 | 163 |
168
- | 15 | 113 |
169
- | 16 | 58 |
170
- | 17 | 57 |
171
- | 18 | 28 |
172
- | 19 | 20 |
173
- | 20 | 7 |
174
- | 21 | 10 |
175
- | 22 | 10 |
176
- | 23 | 2 |
177
- | 24 | 1 |
178
- | 25 | 2 |
179
- | 27 | 2 |
180
- | 29 | 1 |
181
- | 32 | 1 |
182
- | 33 | 2 |
183
- </div>
 
 
 
 
 
 
 
 
184
  </div>
185
 
186
  ---
 
27
 
28
  > **Discord-Dialogues** is a large-scale dataset of anonymized Discord conversations from late spring to early fall 2025 for training and evaluating realistic conversational AI models in a ChatML-friendly format.
29
 
30
+ This dataset contains 7.5 million exchanges spread out over 17 million turns, with more than 145 million words.
31
 
32
  ---
33
 
 
111
  ## Dataset Statistics <span style="font-weight:normal;">(using the [NousResearch/Hermes-3-Llama-3.1-8B tokenizer](https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B))</span>
112
 
113
  <div style="display:flex; gap:20px; align-items:flex-start;">
114
+
115
+ <div>
116
+
117
+ | Metric | Value |
118
+ | ---------------------- | ------------: |
119
+ | Samples (count) | 7,546,294 |
120
+ | Min length (tokens) | 7 |
121
+ | Max length (tokens) | 5,979 |
122
+ | Mean length (tokens) | 33.02 |
123
+ | Median length (tokens) | 29 |
124
+ | Std dev (tokens) | 17.39 |
125
+ | Skew | 26.46 |
126
+ | Kurtosis | 7,487.55 |
127
+ | Total tokens | 249,193,745 |
128
+ | Total characters | 1,291,480,299 |
129
+ | Total words | 145,887,976 |
130
+ | Avg chars per sample | 171.14 |
131
+ | Avg words per sample | 19.33 |
132
+ | Avg chars per word | 8.85 |
133
+ | Tokens per char | 0.19 |
134
+ | Total assistant blocks | 9,341,891 |
135
+
136
+ </div>
137
+
138
+ <div>
139
+
140
+ | Tokens | Count |
141
+ | --------- | --------: |
142
+ | 0–8 | 1 |
143
+ | 8–16 | 110,310 |
144
+ | 16–32 | 4,382,094 |
145
+ | 32–64 | 2,674,780 |
146
+ | 64–128 | 360,401 |
147
+ | 128–256 | 18,083 |
148
+ | 256–384 | 417 |
149
+ | 384–512 | 75 |
150
+ | 512–768 | 78 |
151
+ | 768–1024 | 30 |
152
+ | 1024–2048 | 18 |
153
+ | 2048–4096 | 3 |
154
+
155
+ </div>
156
+
157
+ <div>
158
+
159
+ | Turns | Count |
160
+ | ----- | --------: |
161
+ | 2 | 5,969,540 |
162
+ | 3 | 1,080,526 |
163
+ | 4 | 319,794 |
164
+ | 5 | 102,553 |
165
+ | 6 | 41,246 |
166
+ | 7 | 16,904 |
167
+ | 8 | 7,715 |
168
+ | 9 | 3,691 |
169
+ | 10 | 1,867 |
170
+ | 11 | 1,007 |
171
+ | 12 | 575 |
172
+ | 13 | 334 |
173
+ | 14 | 189 |
174
+ | 15 | 129 |
175
+ | 16 | 67 |
176
+ | 17 | 62 |
177
+ | 18 | 32 |
178
+ | 19 | 21 |
179
+ | 20 | 8 |
180
+ | 21 | 11 |
181
+ | 22 | 11 |
182
+ | 23 | 2 |
183
+ | 24 | 1 |
184
+ | 25 | 3 |
185
+ | 27 | 2 |
186
+ | 29 | 1 |
187
+ | 32 | 1 |
188
+ | 33 | 2 |
189
+
190
+ </div>
191
+
192
  </div>
193
 
194
  ---