Commit
•
c0c68e7
1
Parent(s):
e36d40b
feat: unify flow for generating system prompt and dataset
Browse files
src/distilabel_dataset_generator/sft.py
CHANGED
@@ -290,30 +290,37 @@ More information on distilabel and techniques can be found in the "FAQ" tab. The
|
|
290 |
value=DEFAULT_SYSTEM_PROMPT_DESCRIPTION,
|
291 |
)
|
292 |
|
293 |
-
btn_generate_system_prompt = gr.Button(
|
|
|
|
|
294 |
|
295 |
system_prompt = gr.Textbox(
|
296 |
label="Provide or correct the system prompt",
|
297 |
value=DEFAULT_SYSTEM_PROMPT,
|
298 |
)
|
299 |
|
300 |
-
btn_generate_system_prompt.click(
|
301 |
-
fn=generate_system_prompt,
|
302 |
-
inputs=[dataset_description],
|
303 |
-
outputs=[system_prompt],
|
304 |
-
)
|
305 |
-
|
306 |
btn_generate_sample_dataset = gr.Button(
|
307 |
value="🧪 Generate Sample Dataset of 5 rows and a single turn",
|
308 |
)
|
309 |
|
310 |
table = gr.Dataframe(label="Generated Dataset", wrap=True, value=DEFAULT_DATASET)
|
311 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
312 |
btn_generate_sample_dataset.click(
|
313 |
fn=generate_dataset,
|
314 |
inputs=[system_prompt],
|
315 |
outputs=[table],
|
316 |
)
|
|
|
317 |
btn_login: gr.LoginButton | None = get_login_button()
|
318 |
with gr.Column() as push_to_hub_ui:
|
319 |
with gr.Row(variant="panel"):
|
|
|
290 |
value=DEFAULT_SYSTEM_PROMPT_DESCRIPTION,
|
291 |
)
|
292 |
|
293 |
+
btn_generate_system_prompt = gr.Button(
|
294 |
+
value="🧪 Generate Sytem Prompt and Sample Dataset"
|
295 |
+
)
|
296 |
|
297 |
system_prompt = gr.Textbox(
|
298 |
label="Provide or correct the system prompt",
|
299 |
value=DEFAULT_SYSTEM_PROMPT,
|
300 |
)
|
301 |
|
|
|
|
|
|
|
|
|
|
|
|
|
302 |
btn_generate_sample_dataset = gr.Button(
|
303 |
value="🧪 Generate Sample Dataset of 5 rows and a single turn",
|
304 |
)
|
305 |
|
306 |
table = gr.Dataframe(label="Generated Dataset", wrap=True, value=DEFAULT_DATASET)
|
307 |
|
308 |
+
btn_generate_system_prompt.click(
|
309 |
+
fn=generate_system_prompt,
|
310 |
+
inputs=[dataset_description],
|
311 |
+
outputs=[system_prompt],
|
312 |
+
).then(
|
313 |
+
fn=generate_dataset,
|
314 |
+
inputs=[system_prompt],
|
315 |
+
outputs=[table],
|
316 |
+
)
|
317 |
+
|
318 |
btn_generate_sample_dataset.click(
|
319 |
fn=generate_dataset,
|
320 |
inputs=[system_prompt],
|
321 |
outputs=[table],
|
322 |
)
|
323 |
+
|
324 |
btn_login: gr.LoginButton | None = get_login_button()
|
325 |
with gr.Column() as push_to_hub_ui:
|
326 |
with gr.Row(variant="panel"):
|
src/distilabel_dataset_generator/utils.py
CHANGED
@@ -57,7 +57,8 @@ def list_orgs(token: OAuthToken = None):
|
|
57 |
for entry in data["auth"]["accessToken"]["fineGrained"]["scoped"]
|
58 |
if "repo.write" in entry["permissions"]
|
59 |
]
|
60 |
-
|
|
|
61 |
else:
|
62 |
return []
|
63 |
|
|
|
57 |
for entry in data["auth"]["accessToken"]["fineGrained"]["scoped"]
|
58 |
if "repo.write" in entry["permissions"]
|
59 |
]
|
60 |
+
organisations.append(data["name"])
|
61 |
+
return list(set(organisations))
|
62 |
else:
|
63 |
return []
|
64 |
|