Commit
•
7762f99
1
Parent(s):
3445828
feat/argilla-review (#9)
Browse files- feat: Add Argilla review integration (f1f92f792ab354b38e5297fa293dba8cd384cca4)
- feat: Add blackened tabs (6e87332ce6eabe02888aec2b0eca118149d7824e)
- refactor: Add argilla as first option (35eb40e7ea1d3ebaa1e2f0a349a7f999372c0eda)
- feat: Add validation to check if data exists before pushing/generating (b445efedfc4bf66aa41a3efd61d34f3030bfd41f)
- feat: Add functionality to add records to existing dataset (fc100c211d3e00d90290091a47b2ef5f06f76dfa)
- feat: Add better title (b6ae406e21dd187dbdd5bf57d99f31308037db36)
- refactor: Remove additional info about Argilla (fec287c44220f83c9581b4262a03aecb26fb5b5b)
- app.py +12 -1
- pdm.lock +0 -0
- pyproject.toml +3 -2
- requirements.txt +3 -2
- src/distilabel_dataset_generator/apps/sft.py +314 -32
- src/distilabel_dataset_generator/pipelines/embeddings.py +16 -0
- src/distilabel_dataset_generator/pipelines/sft.py +3 -3
- src/distilabel_dataset_generator/utils.py +14 -0
app.py
CHANGED
@@ -54,6 +54,17 @@ demo = gr.TabbedInterface(
|
|
54 |
margin-bottom: 20px;
|
55 |
}
|
56 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
57 |
</style>
|
58 |
<div class="header-container">
|
59 |
<div class="logo-container">
|
@@ -62,7 +73,7 @@ demo = gr.TabbedInterface(
|
|
62 |
</a>
|
63 |
</div>
|
64 |
<div class="title-container">
|
65 |
-
<h1 style="margin: 0; font-size: 2em;">🧬
|
66 |
<p style="margin: 10px 0 0 0; color: #666; font-size: 1.1em;">Build datasets using natural language</p>
|
67 |
</div>
|
68 |
</div>
|
|
|
54 |
margin-bottom: 20px;
|
55 |
}
|
56 |
}
|
57 |
+
button[role="tab"].selected,
|
58 |
+
button[role="tab"][aria-selected="true"],
|
59 |
+
button[role="tab"][data-tab-id][aria-selected="true"] {
|
60 |
+
background-color: #000000;
|
61 |
+
color: white;
|
62 |
+
border: none;
|
63 |
+
font-size: 16px;
|
64 |
+
font-weight: bold;
|
65 |
+
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2);
|
66 |
+
transition: background-color 0.3s ease, color 0.3s ease;
|
67 |
+
}
|
68 |
</style>
|
69 |
<div class="header-container">
|
70 |
<div class="logo-container">
|
|
|
73 |
</a>
|
74 |
</div>
|
75 |
<div class="title-container">
|
76 |
+
<h1 style="margin: 0; font-size: 2em;">🧬 Synthetic Data Generator</h1>
|
77 |
<p style="margin: 10px 0 0 0; color: #666; font-size: 1.1em;">Build datasets using natural language</p>
|
78 |
</div>
|
79 |
</div>
|
pdm.lock
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
pyproject.toml
CHANGED
@@ -6,11 +6,12 @@ authors = [
|
|
6 |
{name = "davidberenstein1957", email = "[email protected]"},
|
7 |
]
|
8 |
dependencies = [
|
9 |
-
"distilabel[hf-inference-endpoints]
|
10 |
"gradio[oauth]<5,>=4.38",
|
11 |
"transformers>=4.44.2",
|
|
|
12 |
]
|
13 |
-
requires-python = "
|
14 |
readme = "README.md"
|
15 |
license = {text = "apache 2"}
|
16 |
|
|
|
6 |
{name = "davidberenstein1957", email = "[email protected]"},
|
7 |
]
|
8 |
dependencies = [
|
9 |
+
"distilabel[hf-inference-endpoints,argilla]==1.4.0",
|
10 |
"gradio[oauth]<5,>=4.38",
|
11 |
"transformers>=4.44.2",
|
12 |
+
"sentence-transformers>=3.2.0",
|
13 |
]
|
14 |
+
requires-python = "<3.13,>=3.10"
|
15 |
readme = "README.md"
|
16 |
license = {text = "apache 2"}
|
17 |
|
requirements.txt
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
transformers
|
2 |
gradio[oauth]
|
3 |
-
distilabel[hf-inference-endpoints]
|
4 |
-
beautifulsoup4
|
|
|
|
1 |
transformers
|
2 |
gradio[oauth]
|
3 |
+
distilabel[hf-inference-endpoints,argilla]
|
4 |
+
beautifulsoup4
|
5 |
+
sentence-transformers
|
src/distilabel_dataset_generator/apps/sft.py
CHANGED
@@ -1,6 +1,8 @@
|
|
|
|
1 |
import io
|
2 |
-
from typing import Union
|
3 |
|
|
|
4 |
import gradio as gr
|
5 |
import pandas as pd
|
6 |
from datasets import Dataset
|
@@ -8,7 +10,12 @@ from distilabel.distiset import Distiset
|
|
8 |
from distilabel.steps.tasks.text_generation import TextGeneration
|
9 |
from gradio.oauth import OAuthToken
|
10 |
from huggingface_hub import upload_file
|
|
|
11 |
|
|
|
|
|
|
|
|
|
12 |
from src.distilabel_dataset_generator.pipelines.sft import (
|
13 |
DEFAULT_BATCH_SIZE,
|
14 |
DEFAULT_DATASET_DESCRIPTIONS,
|
@@ -21,12 +28,21 @@ from src.distilabel_dataset_generator.pipelines.sft import (
|
|
21 |
get_response_generator,
|
22 |
)
|
23 |
from src.distilabel_dataset_generator.utils import (
|
|
|
24 |
get_login_button,
|
25 |
get_org_dropdown,
|
26 |
swap_visibilty,
|
27 |
)
|
28 |
|
29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
def generate_system_prompt(dataset_description, progress=gr.Progress()):
|
31 |
progress(0.0, desc="Generating system prompt")
|
32 |
if dataset_description in DEFAULT_DATASET_DESCRIPTIONS:
|
@@ -82,7 +98,7 @@ def generate_dataset(
|
|
82 |
num_rows: int = 5,
|
83 |
is_sample: bool = False,
|
84 |
progress=gr.Progress(),
|
85 |
-
):
|
86 |
progress(0.0, desc="(1/2) Generating instructions")
|
87 |
magpie_generator = get_magpie_generator(
|
88 |
num_turns, num_rows, system_prompt, is_sample
|
@@ -191,7 +207,12 @@ def push_to_hub(
|
|
191 |
repo_name: str = None,
|
192 |
oauth_token: Union[OAuthToken, None] = None,
|
193 |
progress=gr.Progress(),
|
194 |
-
):
|
|
|
|
|
|
|
|
|
|
|
195 |
progress(0.1, desc="Setting up dataset")
|
196 |
repo_id = _check_push_to_hub(org_name, repo_name)
|
197 |
distiset = Distiset(
|
@@ -208,7 +229,167 @@ def push_to_hub(
|
|
208 |
create_pr=False,
|
209 |
)
|
210 |
progress(1.0, desc="Dataset pushed to hub")
|
211 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
212 |
|
213 |
|
214 |
def upload_pipeline_code(
|
@@ -313,7 +494,7 @@ with gr.Blocks(
|
|
313 |
# Add a header for the full dataset generation section
|
314 |
gr.Markdown("## Generate full dataset")
|
315 |
gr.Markdown(
|
316 |
-
"Once you're satisfied with the sample, generate a larger dataset and push it to the Hub."
|
317 |
)
|
318 |
|
319 |
with gr.Column() as push_to_hub_ui:
|
@@ -333,27 +514,64 @@ with gr.Blocks(
|
|
333 |
maximum=500,
|
334 |
info="The number of rows in the dataset. Note that you are able to generate more rows at once but that this will take time.",
|
335 |
)
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
|
341 |
-
|
342 |
-
|
343 |
-
|
344 |
-
|
345 |
-
|
346 |
-
|
347 |
-
|
348 |
-
|
349 |
-
|
350 |
-
|
351 |
-
|
352 |
-
|
353 |
-
|
354 |
-
|
355 |
-
|
356 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
357 |
with gr.Row():
|
358 |
final_dataset = gr.Dataframe(
|
359 |
value=DEFAULT_DATASETS[0],
|
@@ -365,7 +583,25 @@ with gr.Blocks(
|
|
365 |
with gr.Row():
|
366 |
success_message = gr.Markdown(visible=False)
|
367 |
|
368 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
369 |
return gr.Markdown(
|
370 |
value=f"""
|
371 |
<div style="padding: 1em; background-color: #e6f3e6; border-radius: 5px; margin-top: 1em;">
|
@@ -378,7 +614,7 @@ with gr.Blocks(
|
|
378 |
</a>
|
379 |
</p>
|
380 |
</div>
|
381 |
-
|
382 |
visible=True,
|
383 |
)
|
384 |
|
@@ -407,8 +643,11 @@ with gr.Blocks(
|
|
407 |
inputs=[sample_dataset],
|
408 |
outputs=[final_dataset],
|
409 |
)
|
410 |
-
|
411 |
-
|
|
|
|
|
|
|
412 |
fn=hide_success_message,
|
413 |
outputs=[success_message],
|
414 |
).then(
|
@@ -418,6 +657,30 @@ with gr.Blocks(
|
|
418 |
show_progress=True,
|
419 |
)
|
420 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
421 |
btn_generate_and_push_to_hub.click(
|
422 |
fn=hide_success_message,
|
423 |
outputs=[success_message],
|
@@ -437,7 +700,7 @@ with gr.Blocks(
|
|
437 |
outputs=[],
|
438 |
show_progress=True,
|
439 |
).success(
|
440 |
-
fn=
|
441 |
inputs=[org_name, repo_name],
|
442 |
outputs=[success_message],
|
443 |
)
|
@@ -456,11 +719,30 @@ with gr.Blocks(
|
|
456 |
outputs=[],
|
457 |
show_progress=True,
|
458 |
).success(
|
459 |
-
fn=
|
460 |
inputs=[org_name, repo_name],
|
461 |
outputs=[success_message],
|
462 |
)
|
463 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
464 |
system_prompt.change(
|
465 |
fn=generate_pipeline_code,
|
466 |
inputs=[system_prompt, num_turns, num_rows],
|
|
|
1 |
+
import ast
|
2 |
import io
|
3 |
+
from typing import Dict, List, Union
|
4 |
|
5 |
+
import argilla as rg
|
6 |
import gradio as gr
|
7 |
import pandas as pd
|
8 |
from datasets import Dataset
|
|
|
10 |
from distilabel.steps.tasks.text_generation import TextGeneration
|
11 |
from gradio.oauth import OAuthToken
|
12 |
from huggingface_hub import upload_file
|
13 |
+
from huggingface_hub.hf_api import HfApi
|
14 |
|
15 |
+
from src.distilabel_dataset_generator.pipelines.embeddings import (
|
16 |
+
get_embeddings,
|
17 |
+
get_sentence_embedding_dimensions,
|
18 |
+
)
|
19 |
from src.distilabel_dataset_generator.pipelines.sft import (
|
20 |
DEFAULT_BATCH_SIZE,
|
21 |
DEFAULT_DATASET_DESCRIPTIONS,
|
|
|
28 |
get_response_generator,
|
29 |
)
|
30 |
from src.distilabel_dataset_generator.utils import (
|
31 |
+
get_argilla_client,
|
32 |
get_login_button,
|
33 |
get_org_dropdown,
|
34 |
swap_visibilty,
|
35 |
)
|
36 |
|
37 |
|
38 |
+
def convert_to_list_of_dicts(messages: str) -> List[Dict[str, str]]:
|
39 |
+
return ast.literal_eval(
|
40 |
+
messages.replace("'user'}", "'user'},")
|
41 |
+
.replace("'system'}", "'system'},")
|
42 |
+
.replace("'assistant'}", "'assistant'},")
|
43 |
+
)
|
44 |
+
|
45 |
+
|
46 |
def generate_system_prompt(dataset_description, progress=gr.Progress()):
|
47 |
progress(0.0, desc="Generating system prompt")
|
48 |
if dataset_description in DEFAULT_DATASET_DESCRIPTIONS:
|
|
|
98 |
num_rows: int = 5,
|
99 |
is_sample: bool = False,
|
100 |
progress=gr.Progress(),
|
101 |
+
) -> pd.DataFrame:
|
102 |
progress(0.0, desc="(1/2) Generating instructions")
|
103 |
magpie_generator = get_magpie_generator(
|
104 |
num_turns, num_rows, system_prompt, is_sample
|
|
|
207 |
repo_name: str = None,
|
208 |
oauth_token: Union[OAuthToken, None] = None,
|
209 |
progress=gr.Progress(),
|
210 |
+
) -> pd.DataFrame:
|
211 |
+
original_dataframe = dataframe.copy(deep=True)
|
212 |
+
if "messages" in dataframe.columns:
|
213 |
+
dataframe["messages"] = dataframe["messages"].apply(
|
214 |
+
lambda x: convert_to_list_of_dicts(x) if isinstance(x, str) else x
|
215 |
+
)
|
216 |
progress(0.1, desc="Setting up dataset")
|
217 |
repo_id = _check_push_to_hub(org_name, repo_name)
|
218 |
distiset = Distiset(
|
|
|
229 |
create_pr=False,
|
230 |
)
|
231 |
progress(1.0, desc="Dataset pushed to hub")
|
232 |
+
return original_dataframe
|
233 |
+
|
234 |
+
|
235 |
+
def push_to_argilla(
|
236 |
+
dataframe: pd.DataFrame,
|
237 |
+
dataset_name: str,
|
238 |
+
oauth_token: Union[OAuthToken, None] = None,
|
239 |
+
progress=gr.Progress(),
|
240 |
+
) -> pd.DataFrame:
|
241 |
+
original_dataframe = dataframe.copy(deep=True)
|
242 |
+
if "messages" in dataframe.columns:
|
243 |
+
dataframe["messages"] = dataframe["messages"].apply(
|
244 |
+
lambda x: convert_to_list_of_dicts(x) if isinstance(x, str) else x
|
245 |
+
)
|
246 |
+
try:
|
247 |
+
progress(0.1, desc="Setting up user and workspace")
|
248 |
+
client = get_argilla_client()
|
249 |
+
hf_user = HfApi().whoami(token=oauth_token.token)["name"]
|
250 |
+
|
251 |
+
# Create user if it doesn't exist
|
252 |
+
rg_user = client.users(username=hf_user)
|
253 |
+
if rg_user is None:
|
254 |
+
rg_user = client.users.add(rg.User(username=hf_user, role="admin"))
|
255 |
+
|
256 |
+
# Create workspace if it doesn't exist
|
257 |
+
workspace = client.workspaces(name=rg_user.username)
|
258 |
+
if workspace is None:
|
259 |
+
workspace = client.workspaces.add(rg.Workspace(name=rg_user.username))
|
260 |
+
workspace.add_user(rg_user)
|
261 |
+
|
262 |
+
if "messages" in dataframe.columns:
|
263 |
+
settings = rg.Settings(
|
264 |
+
fields=[
|
265 |
+
rg.ChatField(
|
266 |
+
name="messages",
|
267 |
+
description="The messages in the conversation",
|
268 |
+
title="Messages",
|
269 |
+
),
|
270 |
+
],
|
271 |
+
questions=[
|
272 |
+
rg.RatingQuestion(
|
273 |
+
name="rating",
|
274 |
+
title="Rating",
|
275 |
+
description="The rating of the conversation",
|
276 |
+
values=list(range(1, 6)),
|
277 |
+
),
|
278 |
+
],
|
279 |
+
metadata=[
|
280 |
+
rg.IntegerMetadataProperty(
|
281 |
+
name="user_message_length", title="User Message Length"
|
282 |
+
),
|
283 |
+
rg.IntegerMetadataProperty(
|
284 |
+
name="assistant_message_length",
|
285 |
+
title="Assistant Message Length",
|
286 |
+
),
|
287 |
+
],
|
288 |
+
vectors=[
|
289 |
+
rg.VectorField(
|
290 |
+
name="messages_embeddings",
|
291 |
+
dimensions=get_sentence_embedding_dimensions(),
|
292 |
+
)
|
293 |
+
],
|
294 |
+
guidelines="Please review the conversation and provide a score for the assistant's response.",
|
295 |
+
)
|
296 |
+
|
297 |
+
dataframe["user_message_length"] = dataframe["messages"].apply(
|
298 |
+
lambda x: sum([len(y["content"]) for y in x if y["role"] == "user"])
|
299 |
+
)
|
300 |
+
dataframe["assistant_message_length"] = dataframe["messages"].apply(
|
301 |
+
lambda x: sum(
|
302 |
+
[len(y["content"]) for y in x if y["role"] == "assistant"]
|
303 |
+
)
|
304 |
+
)
|
305 |
+
dataframe["messages_embeddings"] = get_embeddings(
|
306 |
+
dataframe["messages"].apply(
|
307 |
+
lambda x: " ".join([y["content"] for y in x])
|
308 |
+
)
|
309 |
+
)
|
310 |
+
else:
|
311 |
+
settings = rg.Settings(
|
312 |
+
fields=[
|
313 |
+
rg.TextField(
|
314 |
+
name="system_prompt",
|
315 |
+
title="System Prompt",
|
316 |
+
description="The system prompt used for the conversation",
|
317 |
+
required=False,
|
318 |
+
),
|
319 |
+
rg.TextField(
|
320 |
+
name="prompt",
|
321 |
+
title="Prompt",
|
322 |
+
description="The prompt used for the conversation",
|
323 |
+
),
|
324 |
+
rg.TextField(
|
325 |
+
name="completion",
|
326 |
+
title="Completion",
|
327 |
+
description="The completion from the assistant",
|
328 |
+
),
|
329 |
+
],
|
330 |
+
questions=[
|
331 |
+
rg.RatingQuestion(
|
332 |
+
name="rating",
|
333 |
+
title="Rating",
|
334 |
+
description="The rating of the conversation",
|
335 |
+
values=list(range(1, 6)),
|
336 |
+
),
|
337 |
+
],
|
338 |
+
metadata=[
|
339 |
+
rg.IntegerMetadataProperty(
|
340 |
+
name="prompt_length", title="Prompt Length"
|
341 |
+
),
|
342 |
+
rg.IntegerMetadataProperty(
|
343 |
+
name="completion_length", title="Completion Length"
|
344 |
+
),
|
345 |
+
],
|
346 |
+
vectors=[
|
347 |
+
rg.VectorField(
|
348 |
+
name="prompt_embeddings",
|
349 |
+
dimensions=get_sentence_embedding_dimensions(),
|
350 |
+
)
|
351 |
+
],
|
352 |
+
guidelines="Please review the conversation and correct the prompt and completion where needed.",
|
353 |
+
)
|
354 |
+
dataframe["prompt_length"] = dataframe["prompt"].apply(len)
|
355 |
+
dataframe["completion_length"] = dataframe["completion"].apply(len)
|
356 |
+
dataframe["prompt_embeddings"] = get_embeddings(dataframe["prompt"])
|
357 |
+
|
358 |
+
progress(0.5, desc="Creating dataset")
|
359 |
+
rg_dataset = client.datasets(name=dataset_name, workspace=rg_user.username)
|
360 |
+
if rg_dataset is None:
|
361 |
+
rg_dataset = rg.Dataset(
|
362 |
+
name=dataset_name,
|
363 |
+
workspace=rg_user.username,
|
364 |
+
settings=settings,
|
365 |
+
client=client,
|
366 |
+
)
|
367 |
+
rg_dataset = rg_dataset.create()
|
368 |
+
progress(0.7, desc="Pushing dataset to Argilla")
|
369 |
+
hf_dataset = Dataset.from_pandas(dataframe)
|
370 |
+
rg_dataset.records.log(records=hf_dataset)
|
371 |
+
progress(1.0, desc="Dataset pushed to Argilla")
|
372 |
+
except Exception as e:
|
373 |
+
raise gr.Error(f"Error pushing dataset to Argilla: {e}")
|
374 |
+
return original_dataframe
|
375 |
+
|
376 |
+
|
377 |
+
def validate_argilla_dataset_name(
|
378 |
+
dataset_name: str,
|
379 |
+
final_dataset: pd.DataFrame,
|
380 |
+
add_to_existing_dataset: bool,
|
381 |
+
oauth_token: Union[OAuthToken, None] = None,
|
382 |
+
progress=gr.Progress(),
|
383 |
+
) -> str:
|
384 |
+
progress(0, desc="Validating dataset configuration")
|
385 |
+
hf_user = HfApi().whoami(token=oauth_token.token)["name"]
|
386 |
+
client = get_argilla_client()
|
387 |
+
if dataset_name is None or dataset_name == "":
|
388 |
+
raise gr.Error("Dataset name is required")
|
389 |
+
dataset = client.datasets(name=dataset_name, workspace=hf_user)
|
390 |
+
if dataset and not add_to_existing_dataset:
|
391 |
+
raise gr.Error(f"Dataset {dataset_name} already exists")
|
392 |
+
return final_dataset
|
393 |
|
394 |
|
395 |
def upload_pipeline_code(
|
|
|
494 |
# Add a header for the full dataset generation section
|
495 |
gr.Markdown("## Generate full dataset")
|
496 |
gr.Markdown(
|
497 |
+
"Once you're satisfied with the sample, generate a larger dataset and push it to Argilla or the Hugging Face Hub."
|
498 |
)
|
499 |
|
500 |
with gr.Column() as push_to_hub_ui:
|
|
|
514 |
maximum=500,
|
515 |
info="The number of rows in the dataset. Note that you are able to generate more rows at once but that this will take time.",
|
516 |
)
|
517 |
+
|
518 |
+
with gr.Tab(label="Argilla"):
|
519 |
+
if get_argilla_client():
|
520 |
+
with gr.Row(variant="panel"):
|
521 |
+
dataset_name = gr.Textbox(
|
522 |
+
label="Dataset name",
|
523 |
+
placeholder="dataset_name",
|
524 |
+
value="my-distiset",
|
525 |
+
)
|
526 |
+
add_to_existing_dataset = gr.Checkbox(
|
527 |
+
label="Allow adding records to existing dataset",
|
528 |
+
info="When selected, you do need to ensure the number of turns in the conversation is the same as the number of turns in the existing dataset.",
|
529 |
+
value=False,
|
530 |
+
interactive=True,
|
531 |
+
scale=0.5,
|
532 |
+
)
|
533 |
+
|
534 |
+
with gr.Row(variant="panel"):
|
535 |
+
btn_generate_full_dataset_copy = gr.Button(
|
536 |
+
value="Generate", variant="primary", scale=2
|
537 |
+
)
|
538 |
+
btn_generate_and_push_to_argilla = gr.Button(
|
539 |
+
value="Generate and Push to Argilla",
|
540 |
+
variant="primary",
|
541 |
+
scale=2,
|
542 |
+
)
|
543 |
+
btn_push_to_argilla = gr.Button(
|
544 |
+
value="Push to Argilla", variant="primary", scale=2
|
545 |
+
)
|
546 |
+
else:
|
547 |
+
gr.Markdown(
|
548 |
+
"Please add `ARGILLA_API_URL` and `ARGILLA_API_KEY` to use Argilla."
|
549 |
+
)
|
550 |
+
with gr.Tab("Hugging Face Hub"):
|
551 |
+
with gr.Row(variant="panel"):
|
552 |
+
org_name = get_org_dropdown()
|
553 |
+
repo_name = gr.Textbox(
|
554 |
+
label="Repo name",
|
555 |
+
placeholder="dataset_name",
|
556 |
+
value="my-distiset",
|
557 |
+
)
|
558 |
+
private = gr.Checkbox(
|
559 |
+
label="Private dataset",
|
560 |
+
value=True,
|
561 |
+
interactive=True,
|
562 |
+
scale=0.5,
|
563 |
+
)
|
564 |
+
with gr.Row(variant="panel"):
|
565 |
+
btn_generate_full_dataset = gr.Button(
|
566 |
+
value="Generate", variant="primary", scale=2
|
567 |
+
)
|
568 |
+
btn_generate_and_push_to_hub = gr.Button(
|
569 |
+
value="Generate and Push to Hub", variant="primary", scale=2
|
570 |
+
)
|
571 |
+
btn_push_to_hub = gr.Button(
|
572 |
+
value="Push to Hub", variant="primary", scale=2
|
573 |
+
)
|
574 |
+
|
575 |
with gr.Row():
|
576 |
final_dataset = gr.Dataframe(
|
577 |
value=DEFAULT_DATASETS[0],
|
|
|
583 |
with gr.Row():
|
584 |
success_message = gr.Markdown(visible=False)
|
585 |
|
586 |
+
def show_success_message_argilla():
|
587 |
+
client = get_argilla_client()
|
588 |
+
argilla_api_url = client.api_url
|
589 |
+
return gr.Markdown(
|
590 |
+
value=f"""
|
591 |
+
<div style="padding: 1em; background-color: #e6f3e6; border-radius: 5px; margin-top: 1em;">
|
592 |
+
<h3 style="color: #2e7d32; margin: 0;">Dataset Published Successfully!</h3>
|
593 |
+
<p style="margin-top: 0.5em;">
|
594 |
+
Your dataset is now available at:
|
595 |
+
<a href="{argilla_api_url}" target="_blank" style="color: #1565c0; text-decoration: none;">
|
596 |
+
{argilla_api_url}
|
597 |
+
</a>
|
598 |
+
</p>
|
599 |
+
</div>
|
600 |
+
""",
|
601 |
+
visible=True,
|
602 |
+
)
|
603 |
+
|
604 |
+
def show_success_message_hub(org_name, repo_name):
|
605 |
return gr.Markdown(
|
606 |
value=f"""
|
607 |
<div style="padding: 1em; background-color: #e6f3e6; border-radius: 5px; margin-top: 1em;">
|
|
|
614 |
</a>
|
615 |
</p>
|
616 |
</div>
|
617 |
+
""",
|
618 |
visible=True,
|
619 |
)
|
620 |
|
|
|
643 |
inputs=[sample_dataset],
|
644 |
outputs=[final_dataset],
|
645 |
)
|
646 |
+
gr.on(
|
647 |
+
triggers=[
|
648 |
+
btn_generate_full_dataset.click,
|
649 |
+
btn_generate_full_dataset_copy.click,
|
650 |
+
],
|
651 |
fn=hide_success_message,
|
652 |
outputs=[success_message],
|
653 |
).then(
|
|
|
657 |
show_progress=True,
|
658 |
)
|
659 |
|
660 |
+
btn_generate_and_push_to_argilla.click(
|
661 |
+
fn=validate_argilla_dataset_name,
|
662 |
+
inputs=[dataset_name, final_dataset, add_to_existing_dataset],
|
663 |
+
outputs=[final_dataset],
|
664 |
+
show_progress=True,
|
665 |
+
).success(
|
666 |
+
fn=hide_success_message,
|
667 |
+
outputs=[success_message],
|
668 |
+
).success(
|
669 |
+
fn=generate_dataset,
|
670 |
+
inputs=[system_prompt, num_turns, num_rows],
|
671 |
+
outputs=[final_dataset],
|
672 |
+
show_progress=True,
|
673 |
+
).success(
|
674 |
+
fn=push_to_argilla,
|
675 |
+
inputs=[final_dataset, dataset_name],
|
676 |
+
outputs=[final_dataset],
|
677 |
+
show_progress=True,
|
678 |
+
).success(
|
679 |
+
fn=show_success_message_argilla,
|
680 |
+
inputs=[],
|
681 |
+
outputs=[success_message],
|
682 |
+
)
|
683 |
+
|
684 |
btn_generate_and_push_to_hub.click(
|
685 |
fn=hide_success_message,
|
686 |
outputs=[success_message],
|
|
|
700 |
outputs=[],
|
701 |
show_progress=True,
|
702 |
).success(
|
703 |
+
fn=show_success_message_hub,
|
704 |
inputs=[org_name, repo_name],
|
705 |
outputs=[success_message],
|
706 |
)
|
|
|
719 |
outputs=[],
|
720 |
show_progress=True,
|
721 |
).success(
|
722 |
+
fn=show_success_message_hub,
|
723 |
inputs=[org_name, repo_name],
|
724 |
outputs=[success_message],
|
725 |
)
|
726 |
|
727 |
+
btn_push_to_argilla.click(
|
728 |
+
fn=hide_success_message,
|
729 |
+
outputs=[success_message],
|
730 |
+
).success(
|
731 |
+
fn=validate_argilla_dataset_name,
|
732 |
+
inputs=[dataset_name, final_dataset, add_to_existing_dataset],
|
733 |
+
outputs=[final_dataset],
|
734 |
+
show_progress=True,
|
735 |
+
).success(
|
736 |
+
fn=push_to_argilla,
|
737 |
+
inputs=[final_dataset, dataset_name],
|
738 |
+
outputs=[final_dataset],
|
739 |
+
show_progress=True,
|
740 |
+
).success(
|
741 |
+
fn=show_success_message_argilla,
|
742 |
+
inputs=[],
|
743 |
+
outputs=[success_message],
|
744 |
+
)
|
745 |
+
|
746 |
system_prompt.change(
|
747 |
fn=generate_pipeline_code,
|
748 |
inputs=[system_prompt, num_turns, num_rows],
|
src/distilabel_dataset_generator/pipelines/embeddings.py
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import List
|
2 |
+
|
3 |
+
from sentence_transformers import SentenceTransformer
|
4 |
+
from sentence_transformers.models import StaticEmbedding
|
5 |
+
|
6 |
+
# Initialize a StaticEmbedding module
|
7 |
+
static_embedding = StaticEmbedding.from_model2vec("minishlab/M2V_base_output")
|
8 |
+
model = SentenceTransformer(modules=[static_embedding])
|
9 |
+
|
10 |
+
|
11 |
+
def get_embeddings(texts: List[str]) -> List[List[float]]:
|
12 |
+
return [embedding.tolist() for embedding in model.encode(texts)]
|
13 |
+
|
14 |
+
|
15 |
+
def get_sentence_embedding_dimensions() -> int:
|
16 |
+
return model.get_sentence_embedding_dimension()
|
src/distilabel_dataset_generator/pipelines/sft.py
CHANGED
@@ -189,7 +189,7 @@ with Pipeline(name="sft") as pipeline:
|
|
189 |
tokenizer_id=MODEL,
|
190 |
magpie_pre_query_template="llama3",
|
191 |
generation_kwargs={{
|
192 |
-
"temperature":
|
193 |
"do_sample": True,
|
194 |
"max_new_tokens": 2048,
|
195 |
"stop_sequences": {_STOP_SEQUENCES}
|
@@ -231,7 +231,7 @@ def get_magpie_generator(num_turns, num_rows, system_prompt, is_sample):
|
|
231 |
api_key=_get_next_api_key(),
|
232 |
magpie_pre_query_template="llama3",
|
233 |
generation_kwargs={
|
234 |
-
"temperature":
|
235 |
"do_sample": True,
|
236 |
"max_new_tokens": 256 if is_sample else 512,
|
237 |
"stop_sequences": _STOP_SEQUENCES,
|
@@ -250,7 +250,7 @@ def get_magpie_generator(num_turns, num_rows, system_prompt, is_sample):
|
|
250 |
api_key=_get_next_api_key(),
|
251 |
magpie_pre_query_template="llama3",
|
252 |
generation_kwargs={
|
253 |
-
"temperature":
|
254 |
"do_sample": True,
|
255 |
"max_new_tokens": 256 if is_sample else 1024,
|
256 |
"stop_sequences": _STOP_SEQUENCES,
|
|
|
189 |
tokenizer_id=MODEL,
|
190 |
magpie_pre_query_template="llama3",
|
191 |
generation_kwargs={{
|
192 |
+
"temperature": 1,
|
193 |
"do_sample": True,
|
194 |
"max_new_tokens": 2048,
|
195 |
"stop_sequences": {_STOP_SEQUENCES}
|
|
|
231 |
api_key=_get_next_api_key(),
|
232 |
magpie_pre_query_template="llama3",
|
233 |
generation_kwargs={
|
234 |
+
"temperature": 1,
|
235 |
"do_sample": True,
|
236 |
"max_new_tokens": 256 if is_sample else 512,
|
237 |
"stop_sequences": _STOP_SEQUENCES,
|
|
|
250 |
api_key=_get_next_api_key(),
|
251 |
magpie_pre_query_template="llama3",
|
252 |
generation_kwargs={
|
253 |
+
"temperature": 1,
|
254 |
"do_sample": True,
|
255 |
"max_new_tokens": 256 if is_sample else 1024,
|
256 |
"stop_sequences": _STOP_SEQUENCES,
|
src/distilabel_dataset_generator/utils.py
CHANGED
@@ -1,5 +1,7 @@
|
|
1 |
import os
|
|
|
2 |
|
|
|
3 |
import gradio as gr
|
4 |
from gradio.oauth import (
|
5 |
OAUTH_CLIENT_ID,
|
@@ -81,3 +83,15 @@ def swap_visibilty(oauth_token: OAuthToken = None):
|
|
81 |
return gr.update(elem_classes=["main_ui_logged_in"])
|
82 |
else:
|
83 |
return gr.update(elem_classes=["main_ui_logged_out"])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import os
|
2 |
+
from typing import Union
|
3 |
|
4 |
+
import argilla as rg
|
5 |
import gradio as gr
|
6 |
from gradio.oauth import (
|
7 |
OAUTH_CLIENT_ID,
|
|
|
83 |
return gr.update(elem_classes=["main_ui_logged_in"])
|
84 |
else:
|
85 |
return gr.update(elem_classes=["main_ui_logged_out"])
|
86 |
+
|
87 |
+
|
88 |
+
def get_argilla_client() -> Union[rg.Argilla, None]:
|
89 |
+
try:
|
90 |
+
return rg.Argilla(
|
91 |
+
api_url=os.getenv("ARGILLA_API_URL_SDG_REVIEWER")
|
92 |
+
or os.getenv("ARGILLA_API_URL"),
|
93 |
+
api_key=os.getenv("ARGILLA_API_KEY_SDG_REVIEWER")
|
94 |
+
or os.getenv("ARGILLA_API_KEY"),
|
95 |
+
)
|
96 |
+
except Exception:
|
97 |
+
return None
|