awoo
Browse filesSigned-off-by: Balazs Horvath <[email protected]>
- README.md +56 -27
- dataset_tools/Count Tokens in Sample Prompts.ipynb +193 -0
README.md
CHANGED
@@ -30,17 +30,18 @@ The Yiff Toolkit is a comprehensive set of tools designed to enhance your creati
|
|
30 |
- [Pony Training](#pony-training)
|
31 |
- [Download Pony in Diffusers Format](#download-pony-in-diffusers-format)
|
32 |
- [Sample Prompt File](#sample-prompt-file)
|
33 |
-
- [
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
|
|
44 |
- [`--sdpa` or `--xformers` or `--mem_eff_attn`](#--sdpa-or---xformers-or---mem_eff_attn)
|
45 |
- [`--sample_prompts` and `--sample_sampler` and `--sample_every_n_steps`](#--sample_prompts-and---sample_sampler-and---sample_every_n_steps)
|
46 |
- [CosXL Training](#cosxl-training)
|
@@ -141,18 +142,24 @@ A sample prompt file is used during training to sample images. A sample prompt f
|
|
141 |
|
142 |
```py
|
143 |
# anthro female kindred
|
144 |
-
score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, female anthro kindred presenting, white pillow, bedroom, looking at viewer, detailed background, amazing_background, scenery porn, realistic, photo
|
145 |
# anthro female wolf
|
146 |
-
score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, anthro female wolf, sexy pose, standing, gray fur, brown fur, canine pussy, black nose,
|
147 |
-
# anthro male fox
|
148 |
-
score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, anthro male fox, glowing yellow eyes, night, crescent moon, tibetan necklace, gold bracers, blue and gold adorned loincloth, canine genitalia, knot, amazing_background, scenery porn, white marble ruins in the background, realistic, photo, photo (medium), photography (artwork) --n low quality, worst quality --w 1024 --h 1024 --d 1 --l 6.0 --s 40
|
149 |
```
|
150 |
|
151 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
152 |
|
153 |
If you are running running out of RAM like I do with 2 GPUs and a really fat model, this option will help you save a bit of it and might get you out of OOM hell.
|
154 |
|
155 |
-
|
|
|
|
|
156 |
|
157 |
The directory containing the checkpoint you just downloaded. I recommend closing the path if you are using a local model with a `/`.
|
158 |
|
@@ -160,7 +167,9 @@ The directory containing the checkpoint you just downloaded. I recommend closing
|
|
160 |
--pretrained_model_name_or_path="/ponydiffusers/" \
|
161 |
```
|
162 |
|
163 |
-
|
|
|
|
|
164 |
|
165 |
This is where all the saved epochs or steps will be saved, including the last one. If y
|
166 |
|
@@ -168,7 +177,9 @@ This is where all the saved epochs or steps will be saved, including the last on
|
|
168 |
--output_dir="/output_dir" \
|
169 |
```
|
170 |
|
171 |
-
|
|
|
|
|
172 |
|
173 |
The directory containing the dataset. We prepared this earlier together.
|
174 |
|
@@ -176,7 +187,9 @@ The directory containing the dataset. We prepared this earlier together.
|
|
176 |
--train_data_dir="/training_dir" \
|
177 |
```
|
178 |
|
179 |
-
|
|
|
|
|
180 |
|
181 |
Always set this to match the model's resolution, which in Pony's case it is 1024x1024. If you can't fit into the VRAM, you can decrease it to `512,512` as a last resort.
|
182 |
|
@@ -184,7 +197,9 @@ Always set this to match the model's resolution, which in Pony's case it is 1024
|
|
184 |
--resolution="1024,1024" \
|
185 |
```
|
186 |
|
187 |
-
|
|
|
|
|
188 |
|
189 |
β οΈ
|
190 |
|
@@ -192,7 +207,9 @@ Always set this to match the model's resolution, which in Pony's case it is 1024
|
|
192 |
--enable_bucket \
|
193 |
```
|
194 |
|
195 |
-
|
|
|
|
|
196 |
|
197 |
β οΈ
|
198 |
|
@@ -201,15 +218,19 @@ Always set this to match the model's resolution, which in Pony's case it is 1024
|
|
201 |
--max_bucket_reso=1024 \
|
202 |
```
|
203 |
|
204 |
-
|
|
|
|
|
205 |
|
206 |
-
The default optimizer is `AdamW` and there are a bunch of them added every month or so, therefore I'm not listing
|
207 |
|
208 |
```py
|
209 |
--optimizer_type="AdamW" \
|
210 |
```
|
211 |
|
212 |
-
|
|
|
|
|
213 |
|
214 |
Repeats the dataset when training with captions, by default it is set to `1` so we'll set this to `0` with:
|
215 |
|
@@ -217,7 +238,9 @@ Repeats the dataset when training with captions, by default it is set to `1` so
|
|
217 |
--dataset_repeats=0 \
|
218 |
```
|
219 |
|
220 |
-
|
|
|
|
|
221 |
|
222 |
Specify the number of steps or epochs to train. If both `--max_train_steps` and `--max_train_epochs` are specified, the number of epochs takes precedence.
|
223 |
|
@@ -225,7 +248,9 @@ Specify the number of steps or epochs to train. If both `--max_train_steps` and
|
|
225 |
--max_train_steps=500 \
|
226 |
```
|
227 |
|
228 |
-
|
|
|
|
|
229 |
|
230 |
Shuffles the captions set by `--caption_separator`, it is a comma `,` by default which will work perfectly for our case since our captions look like this:
|
231 |
|
@@ -233,6 +258,10 @@ Shuffles the captions set by `--caption_separator`, it is a comma `,` by default
|
|
233 |
|
234 |
As you can tell, I have separated the caption part not just the tags with a `,` to make sure everything gets shuffled. I'm at this point pretty certain this is beneficial especially when your caption file contains more than 77 tokens.
|
235 |
|
|
|
|
|
|
|
|
|
236 |
#### `--sdpa` or `--xformers` or `--mem_eff_attn`
|
237 |
|
238 |
The choice between `--xformers` or `--mem_eff_attn` and `--spda` will depend on your GPU. You can benchmark it by repeating a training with them!
|
|
|
30 |
- [Pony Training](#pony-training)
|
31 |
- [Download Pony in Diffusers Format](#download-pony-in-diffusers-format)
|
32 |
- [Sample Prompt File](#sample-prompt-file)
|
33 |
+
- [Training Commands](#training-commands)
|
34 |
+
- [`--lowram`](#--lowram)
|
35 |
+
- [`--pretrained_model_name_or_path`](#--pretrained_model_name_or_path)
|
36 |
+
- [`--output_dir`](#--output_dir)
|
37 |
+
- [`--train_data_dir`](#--train_data_dir)
|
38 |
+
- [`--resolution`](#--resolution)
|
39 |
+
- [`--enable_bucket`](#--enable_bucket)
|
40 |
+
- [`--min_bucket_reso` and `--max_bucket_reso`](#--min_bucket_reso-and---max_bucket_reso)
|
41 |
+
- [`--optimizer_type`](#--optimizer_type)
|
42 |
+
- [`--dataset_repeats`](#--dataset_repeats)
|
43 |
+
- [`--max_train_steps`](#--max_train_steps)
|
44 |
+
- [`--shuffle_caption`](#--shuffle_caption)
|
45 |
- [`--sdpa` or `--xformers` or `--mem_eff_attn`](#--sdpa-or---xformers-or---mem_eff_attn)
|
46 |
- [`--sample_prompts` and `--sample_sampler` and `--sample_every_n_steps`](#--sample_prompts-and---sample_sampler-and---sample_every_n_steps)
|
47 |
- [CosXL Training](#cosxl-training)
|
|
|
142 |
|
143 |
```py
|
144 |
# anthro female kindred
|
145 |
+
score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, female anthro kindred presenting, white pillow, bedroom, looking at viewer, detailed background, amazing_background, scenery porn, realistic, photo --n low quality, worst quality, blurred background, blurry, simple background --w 1024 --h 1024 --d 1 --l 6.0 --s 40
|
146 |
# anthro female wolf
|
147 |
+
score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, anthro female wolf, sexy pose, standing, gray fur, brown fur, canine pussy, black nose, blue eyes, pink areola, pink nipples, detailed background, amazing_background, realistic, photo --n low quality, worst quality, blurred background, blurry, simple background --w 1024 --h 1024 --d 1 --l 6.0 --s 40
|
|
|
|
|
148 |
```
|
149 |
|
150 |
+
Please note that sample prompts should not exceed 77 tokens, you can use
|
151 |
+
|
152 |
+
---
|
153 |
+
|
154 |
+
#### Training Commands
|
155 |
+
|
156 |
+
##### `--lowram`
|
157 |
|
158 |
If you are running running out of RAM like I do with 2 GPUs and a really fat model, this option will help you save a bit of it and might get you out of OOM hell.
|
159 |
|
160 |
+
---
|
161 |
+
|
162 |
+
##### `--pretrained_model_name_or_path`
|
163 |
|
164 |
The directory containing the checkpoint you just downloaded. I recommend closing the path if you are using a local model with a `/`.
|
165 |
|
|
|
167 |
--pretrained_model_name_or_path="/ponydiffusers/" \
|
168 |
```
|
169 |
|
170 |
+
---
|
171 |
+
|
172 |
+
##### `--output_dir`
|
173 |
|
174 |
This is where all the saved epochs or steps will be saved, including the last one. If y
|
175 |
|
|
|
177 |
--output_dir="/output_dir" \
|
178 |
```
|
179 |
|
180 |
+
---
|
181 |
+
|
182 |
+
##### `--train_data_dir`
|
183 |
|
184 |
The directory containing the dataset. We prepared this earlier together.
|
185 |
|
|
|
187 |
--train_data_dir="/training_dir" \
|
188 |
```
|
189 |
|
190 |
+
---
|
191 |
+
|
192 |
+
##### `--resolution`
|
193 |
|
194 |
Always set this to match the model's resolution, which in Pony's case it is 1024x1024. If you can't fit into the VRAM, you can decrease it to `512,512` as a last resort.
|
195 |
|
|
|
197 |
--resolution="1024,1024" \
|
198 |
```
|
199 |
|
200 |
+
---
|
201 |
+
|
202 |
+
##### `--enable_bucket`
|
203 |
|
204 |
β οΈ
|
205 |
|
|
|
207 |
--enable_bucket \
|
208 |
```
|
209 |
|
210 |
+
---
|
211 |
+
|
212 |
+
##### `--min_bucket_reso` and `--max_bucket_reso`
|
213 |
|
214 |
β οΈ
|
215 |
|
|
|
218 |
--max_bucket_reso=1024 \
|
219 |
```
|
220 |
|
221 |
+
---
|
222 |
+
|
223 |
+
##### `--optimizer_type`
|
224 |
|
225 |
+
The default optimizer is `AdamW` and there are a bunch of them added every month or so, therefore I'm not listing them all, you can find the list if you really want, but `AdamW` is the best as of this writing so we use that!
|
226 |
|
227 |
```py
|
228 |
--optimizer_type="AdamW" \
|
229 |
```
|
230 |
|
231 |
+
---
|
232 |
+
|
233 |
+
##### `--dataset_repeats`
|
234 |
|
235 |
Repeats the dataset when training with captions, by default it is set to `1` so we'll set this to `0` with:
|
236 |
|
|
|
238 |
--dataset_repeats=0 \
|
239 |
```
|
240 |
|
241 |
+
---
|
242 |
+
|
243 |
+
##### `--max_train_steps`
|
244 |
|
245 |
Specify the number of steps or epochs to train. If both `--max_train_steps` and `--max_train_epochs` are specified, the number of epochs takes precedence.
|
246 |
|
|
|
248 |
--max_train_steps=500 \
|
249 |
```
|
250 |
|
251 |
+
---
|
252 |
+
|
253 |
+
##### `--shuffle_caption`
|
254 |
|
255 |
Shuffles the captions set by `--caption_separator`, it is a comma `,` by default which will work perfectly for our case since our captions look like this:
|
256 |
|
|
|
258 |
|
259 |
As you can tell, I have separated the caption part not just the tags with a `,` to make sure everything gets shuffled. I'm at this point pretty certain this is beneficial especially when your caption file contains more than 77 tokens.
|
260 |
|
261 |
+
NOTE: `--cache_text_encoder_outputs` and `--cache_text_encoder_outputs_to_disk` can't be used together with `--shuffle_caption`. Both of these aim to reduce VRAM usage, you will need to decide between these yourself!
|
262 |
+
|
263 |
+
---
|
264 |
+
|
265 |
#### `--sdpa` or `--xformers` or `--mem_eff_attn`
|
266 |
|
267 |
The choice between `--xformers` or `--mem_eff_attn` and `--spda` will depend on your GPU. You can benchmark it by repeating a training with them!
|
dataset_tools/Count Tokens in Sample Prompts.ipynb
ADDED
@@ -0,0 +1,193 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": 4,
|
6 |
+
"metadata": {},
|
7 |
+
"outputs": [
|
8 |
+
{
|
9 |
+
"data": {
|
10 |
+
"text/html": [
|
11 |
+
"<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-style: italic\"> Prompt Analysis </span>\n",
|
12 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
13 |
+
"β<span style=\"font-weight: bold\"> Prompt Type </span>β<span style=\"font-weight: bold\"> Prompt </span>β<span style=\"font-weight: bold\"> Token Count </span>β\n",
|
14 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
15 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 57 β\n",
|
16 |
+
"β β female anthro kindred presenting, white pillow, bedroom, looking at viewer, β β\n",
|
17 |
+
"β β detailed background, amazing_background, scenery porn, realistic, photo β β\n",
|
18 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
19 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n",
|
20 |
+
"</pre>\n"
|
21 |
+
],
|
22 |
+
"text/plain": [
|
23 |
+
"\u001b[3m Prompt Analysis \u001b[0m\n",
|
24 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
25 |
+
"β\u001b[1m \u001b[0m\u001b[1mPrompt Type\u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mPrompt \u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mToken Count\u001b[0m\u001b[1m \u001b[0mβ\n",
|
26 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
27 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 57 β\n",
|
28 |
+
"β β female anthro kindred presenting, white pillow, bedroom, looking at viewer, β β\n",
|
29 |
+
"β β detailed background, amazing_background, scenery porn, realistic, photo β β\n",
|
30 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
31 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n"
|
32 |
+
]
|
33 |
+
},
|
34 |
+
"metadata": {},
|
35 |
+
"output_type": "display_data"
|
36 |
+
},
|
37 |
+
{
|
38 |
+
"data": {
|
39 |
+
"text/html": [
|
40 |
+
"<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-style: italic\"> Prompt Analysis </span>\n",
|
41 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
42 |
+
"β<span style=\"font-weight: bold\"> Prompt Type </span>β<span style=\"font-weight: bold\"> Prompt </span>β<span style=\"font-weight: bold\"> Token Count </span>β\n",
|
43 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
44 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 70 β\n",
|
45 |
+
"β β anthro female wolf, sexy pose, standing, gray fur, brown fur, canine pussy, black β β\n",
|
46 |
+
"β β nose, blue eyes, pink areola, pink nipples, detailed background, β β\n",
|
47 |
+
"β β amazing_background, realistic, photo β β\n",
|
48 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
49 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n",
|
50 |
+
"</pre>\n"
|
51 |
+
],
|
52 |
+
"text/plain": [
|
53 |
+
"\u001b[3m Prompt Analysis \u001b[0m\n",
|
54 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
55 |
+
"β\u001b[1m \u001b[0m\u001b[1mPrompt Type\u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mPrompt \u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mToken Count\u001b[0m\u001b[1m \u001b[0mβ\n",
|
56 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
57 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 70 β\n",
|
58 |
+
"β β anthro female wolf, sexy pose, standing, gray fur, brown fur, canine pussy, black β β\n",
|
59 |
+
"β β nose, blue eyes, pink areola, pink nipples, detailed background, β β\n",
|
60 |
+
"β β amazing_background, realistic, photo β β\n",
|
61 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
62 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n"
|
63 |
+
]
|
64 |
+
},
|
65 |
+
"metadata": {},
|
66 |
+
"output_type": "display_data"
|
67 |
+
},
|
68 |
+
{
|
69 |
+
"data": {
|
70 |
+
"text/html": [
|
71 |
+
"<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-style: italic\"> Prompt Analysis </span>\n",
|
72 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
73 |
+
"β<span style=\"font-weight: bold\"> Prompt Type </span>β<span style=\"font-weight: bold\"> Prompt </span>β<span style=\"font-weight: bold\"> Token Count </span>β\n",
|
74 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
75 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 73 β\n",
|
76 |
+
"β β anthro male fox, glowing yellow eyes, night, crescent moon, gold bracers and β β\n",
|
77 |
+
"β β necklace, loincloth, canine genitalia, knot, amazing_background, scenery porn, β β\n",
|
78 |
+
"β β white marble ruins, realistic, photo β β\n",
|
79 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
80 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n",
|
81 |
+
"</pre>\n"
|
82 |
+
],
|
83 |
+
"text/plain": [
|
84 |
+
"\u001b[3m Prompt Analysis \u001b[0m\n",
|
85 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
86 |
+
"β\u001b[1m \u001b[0m\u001b[1mPrompt Type\u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mPrompt \u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mToken Count\u001b[0m\u001b[1m \u001b[0mβ\n",
|
87 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
88 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_explicit, source_furry, solo, β 73 β\n",
|
89 |
+
"β β anthro male fox, glowing yellow eyes, night, crescent moon, gold bracers and β β\n",
|
90 |
+
"β β necklace, loincloth, canine genitalia, knot, amazing_background, scenery porn, β β\n",
|
91 |
+
"β β white marble ruins, realistic, photo β β\n",
|
92 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
93 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n"
|
94 |
+
]
|
95 |
+
},
|
96 |
+
"metadata": {},
|
97 |
+
"output_type": "display_data"
|
98 |
+
},
|
99 |
+
{
|
100 |
+
"data": {
|
101 |
+
"text/html": [
|
102 |
+
"<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\"><span style=\"font-style: italic\"> Prompt Analysis </span>\n",
|
103 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
104 |
+
"β<span style=\"font-weight: bold\"> Prompt Type </span>β<span style=\"font-weight: bold\"> Prompt </span>β<span style=\"font-weight: bold\"> Token Count </span>β\n",
|
105 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
106 |
+
"β Positive οΏ½οΏ½οΏ½ score_9, score_8_up, score_7_up, score_6_up, rating_safe, source_furry, solo, β 74 β\n",
|
107 |
+
"β β full-length portrait, anthro female red panda, detailed background, β β\n",
|
108 |
+
"β β amazing_background, pussy, scenery porn, photo, realistic, looking at viewer, on β β\n",
|
109 |
+
"β β back, sexy pose, humanoid hands, claws, pink areola, pink nipples β β\n",
|
110 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
111 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n",
|
112 |
+
"</pre>\n"
|
113 |
+
],
|
114 |
+
"text/plain": [
|
115 |
+
"\u001b[3m Prompt Analysis \u001b[0m\n",
|
116 |
+
"βββββββββββββββ³ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ³ββββββββββββββ\n",
|
117 |
+
"β\u001b[1m \u001b[0m\u001b[1mPrompt Type\u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mPrompt \u001b[0m\u001b[1m \u001b[0mβ\u001b[1m \u001b[0m\u001b[1mToken Count\u001b[0m\u001b[1m \u001b[0mβ\n",
|
118 |
+
"β‘ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ©\n",
|
119 |
+
"β Positive β score_9, score_8_up, score_7_up, score_6_up, rating_safe, source_furry, solo, β 74 β\n",
|
120 |
+
"β β full-length portrait, anthro female red panda, detailed background, β β\n",
|
121 |
+
"β β amazing_background, pussy, scenery porn, photo, realistic, looking at viewer, on β β\n",
|
122 |
+
"β β back, sexy pose, humanoid hands, claws, pink areola, pink nipples β β\n",
|
123 |
+
"β Negative β low quality, worst quality, blurred background, blurry, simple background β 13 β\n",
|
124 |
+
"βββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββ\n"
|
125 |
+
]
|
126 |
+
},
|
127 |
+
"metadata": {},
|
128 |
+
"output_type": "display_data"
|
129 |
+
}
|
130 |
+
],
|
131 |
+
"source": [
|
132 |
+
"import tiktoken\n",
|
133 |
+
"from rich.console import Console\n",
|
134 |
+
"from rich.table import Table\n",
|
135 |
+
"\n",
|
136 |
+
"def count_tokens(text):\n",
|
137 |
+
" enc = tiktoken.get_encoding(\"cl100k_base\")\n",
|
138 |
+
" tokens = enc.encode(text)\n",
|
139 |
+
" return len(tokens)\n",
|
140 |
+
"\n",
|
141 |
+
"console = Console()\n",
|
142 |
+
"\n",
|
143 |
+
"with open(\"C:\\\\Users\\\\kade\\\\Desktop\\\\training_dir_staging\\\\sample-prompts.txt\", \"r\") as file:\n",
|
144 |
+
" lines = file.readlines()\n",
|
145 |
+
"\n",
|
146 |
+
"for line in lines:\n",
|
147 |
+
" if line.startswith(\"#\"):\n",
|
148 |
+
" continue\n",
|
149 |
+
"\n",
|
150 |
+
" parts = line.split(\"--n\")\n",
|
151 |
+
" positive_prompt = parts[0].strip()\n",
|
152 |
+
" negative_prompt = parts[1].strip().split(\" --\")[0]\n",
|
153 |
+
"\n",
|
154 |
+
" positive_token_count = count_tokens(positive_prompt)\n",
|
155 |
+
" negative_token_count = count_tokens(negative_prompt)\n",
|
156 |
+
"\n",
|
157 |
+
" table = Table(title=\"Prompt Analysis\")\n",
|
158 |
+
" table.add_column(\"Prompt Type\", justify=\"left\")\n",
|
159 |
+
" table.add_column(\"Prompt\", justify=\"left\")\n",
|
160 |
+
" table.add_column(\"Token Count\", justify=\"right\")\n",
|
161 |
+
"\n",
|
162 |
+
" table.add_row(\"Positive\", positive_prompt, str(positive_token_count))\n",
|
163 |
+
" table.add_row(\"Negative\", negative_prompt, str(negative_token_count))\n",
|
164 |
+
"\n",
|
165 |
+
" console.print(table)\n",
|
166 |
+
"\n",
|
167 |
+
" if positive_token_count > 77:\n",
|
168 |
+
" console.print(f\"[bold red]Warning: Positive prompt token count exceeds 75.[/bold red]\")"
|
169 |
+
]
|
170 |
+
}
|
171 |
+
],
|
172 |
+
"metadata": {
|
173 |
+
"kernelspec": {
|
174 |
+
"display_name": "base",
|
175 |
+
"language": "python",
|
176 |
+
"name": "python3"
|
177 |
+
},
|
178 |
+
"language_info": {
|
179 |
+
"codemirror_mode": {
|
180 |
+
"name": "ipython",
|
181 |
+
"version": 3
|
182 |
+
},
|
183 |
+
"file_extension": ".py",
|
184 |
+
"mimetype": "text/x-python",
|
185 |
+
"name": "python",
|
186 |
+
"nbconvert_exporter": "python",
|
187 |
+
"pygments_lexer": "ipython3",
|
188 |
+
"version": "3.12.2"
|
189 |
+
}
|
190 |
+
},
|
191 |
+
"nbformat": 4,
|
192 |
+
"nbformat_minor": 2
|
193 |
+
}
|