Commit
·
f230d42
1
Parent(s):
545e114
fix correct `OauthToken.token` attribute getter
Browse files
src/synthetic_dataset_generator/apps/base.py
CHANGED
|
@@ -1,19 +1,13 @@
|
|
| 1 |
import io
|
| 2 |
import uuid
|
| 3 |
-
from typing import
|
| 4 |
|
| 5 |
import argilla as rg
|
| 6 |
import gradio as gr
|
| 7 |
-
import pandas as pd
|
| 8 |
-
from datasets import ClassLabel, Dataset, Features, Sequence, Value
|
| 9 |
-
from distilabel.distiset import Distiset
|
| 10 |
from gradio import OAuthToken
|
| 11 |
from huggingface_hub import HfApi, upload_file
|
| 12 |
|
| 13 |
-
from synthetic_dataset_generator.
|
| 14 |
-
from synthetic_dataset_generator.utils import (
|
| 15 |
-
get_argilla_client,
|
| 16 |
-
)
|
| 17 |
|
| 18 |
|
| 19 |
def validate_argilla_user_workspace_dataset(
|
|
@@ -52,7 +46,7 @@ def push_pipeline_code_to_hub(
|
|
| 52 |
oauth_token: Union[OAuthToken, None] = None,
|
| 53 |
progress=gr.Progress(),
|
| 54 |
):
|
| 55 |
-
repo_id = validate_push_to_hub(org_name, repo_name)
|
| 56 |
progress(0.1, desc="Uploading pipeline code")
|
| 57 |
with io.BytesIO(pipeline_code.encode("utf-8")) as f:
|
| 58 |
upload_file(
|
|
|
|
| 1 |
import io
|
| 2 |
import uuid
|
| 3 |
+
from typing import Union
|
| 4 |
|
| 5 |
import argilla as rg
|
| 6 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
| 7 |
from gradio import OAuthToken
|
| 8 |
from huggingface_hub import HfApi, upload_file
|
| 9 |
|
| 10 |
+
from synthetic_dataset_generator.utils import get_argilla_client
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
|
| 13 |
def validate_argilla_user_workspace_dataset(
|
|
|
|
| 46 |
oauth_token: Union[OAuthToken, None] = None,
|
| 47 |
progress=gr.Progress(),
|
| 48 |
):
|
| 49 |
+
repo_id: str | None = validate_push_to_hub(org_name, repo_name)
|
| 50 |
progress(0.1, desc="Uploading pipeline code")
|
| 51 |
with io.BytesIO(pipeline_code.encode("utf-8")) as f:
|
| 52 |
upload_file(
|
src/synthetic_dataset_generator/apps/eval.py
CHANGED
|
@@ -350,7 +350,7 @@ def push_dataset_to_hub(
|
|
| 350 |
dataframe: pd.DataFrame,
|
| 351 |
org_name: str,
|
| 352 |
repo_name: str,
|
| 353 |
-
oauth_token,
|
| 354 |
private: bool,
|
| 355 |
pipeline_code: str,
|
| 356 |
):
|
|
@@ -363,7 +363,7 @@ def push_dataset_to_hub(
|
|
| 363 |
token=oauth_token.token,
|
| 364 |
create_pr=False,
|
| 365 |
)
|
| 366 |
-
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token
|
| 367 |
|
| 368 |
|
| 369 |
def push_dataset(
|
|
|
|
| 350 |
dataframe: pd.DataFrame,
|
| 351 |
org_name: str,
|
| 352 |
repo_name: str,
|
| 353 |
+
oauth_token: Union[gr.OAuthToken, None],
|
| 354 |
private: bool,
|
| 355 |
pipeline_code: str,
|
| 356 |
):
|
|
|
|
| 363 |
token=oauth_token.token,
|
| 364 |
create_pr=False,
|
| 365 |
)
|
| 366 |
+
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token)
|
| 367 |
|
| 368 |
|
| 369 |
def push_dataset(
|
src/synthetic_dataset_generator/apps/sft.py
CHANGED
|
@@ -207,7 +207,7 @@ def push_dataset_to_hub(
|
|
| 207 |
dataframe: pd.DataFrame,
|
| 208 |
org_name: str,
|
| 209 |
repo_name: str,
|
| 210 |
-
oauth_token: gr.OAuthToken,
|
| 211 |
private: bool,
|
| 212 |
pipeline_code: str,
|
| 213 |
):
|
|
@@ -222,7 +222,7 @@ def push_dataset_to_hub(
|
|
| 222 |
token=oauth_token.token,
|
| 223 |
create_pr=False,
|
| 224 |
)
|
| 225 |
-
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token
|
| 226 |
return original_dataframe
|
| 227 |
|
| 228 |
|
|
|
|
| 207 |
dataframe: pd.DataFrame,
|
| 208 |
org_name: str,
|
| 209 |
repo_name: str,
|
| 210 |
+
oauth_token: Union[gr.OAuthToken, None],
|
| 211 |
private: bool,
|
| 212 |
pipeline_code: str,
|
| 213 |
):
|
|
|
|
| 222 |
token=oauth_token.token,
|
| 223 |
create_pr=False,
|
| 224 |
)
|
| 225 |
+
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token)
|
| 226 |
return original_dataframe
|
| 227 |
|
| 228 |
|
src/synthetic_dataset_generator/apps/textcat.py
CHANGED
|
@@ -197,7 +197,7 @@ def push_dataset_to_hub(
|
|
| 197 |
token=oauth_token.token,
|
| 198 |
create_pr=False,
|
| 199 |
)
|
| 200 |
-
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token
|
| 201 |
|
| 202 |
|
| 203 |
def push_dataset(
|
|
|
|
| 197 |
token=oauth_token.token,
|
| 198 |
create_pr=False,
|
| 199 |
)
|
| 200 |
+
push_pipeline_code_to_hub(pipeline_code, org_name, repo_name, oauth_token)
|
| 201 |
|
| 202 |
|
| 203 |
def push_dataset(
|