feat: 🎸 use main instead of master to load datasets (#479)

fixes #438
huggingface · Jul 26, 2022 · 519cf70 · 519cf70
1 parent ca2498a
commit 519cf70
Show file tree

Hide file tree

Showing 4 changed files with 7 additions and 7 deletions.
diff --git a/chart/values.yaml b/chart/values.yaml
@@ -114,7 +114,7 @@ worker:
     # Directory of the "datasets" library cache (for the datasets, not the modules)
     cacheDirectory: "/cache"
     # Git reference for the canonical datasets on https://github.com/huggingface/datasets
-    datasetsRevision: "master"
+    datasetsRevision: "main"
     # User Access Token (see https://huggingface.co/settings/token, only the `read` role is required)
     hfToken: ""
     # Log level
@@ -158,7 +158,7 @@ worker:
     # Directory of the "datasets" library cache (for the datasets, not the modules)
     cacheDirectory: "/cache"
     # Git reference for the canonical datasets on https://github.com/huggingface/datasets
-    datasetsRevision: "master"
+    datasetsRevision: "main"
     # User Access Token (see https://huggingface.co/settings/token, only the `read` role is required)
     hfToken: ""
     # Log level
@@ -202,7 +202,7 @@ worker:
     # Directory of the "datasets" library cache (for the datasets, not the modules)
     cacheDirectory: "/cache"
     # Git reference for the canonical datasets on https://github.com/huggingface/datasets
-    datasetsRevision: "master"
+    datasetsRevision: "main"
     # Log level
     logLevel: "INFO"
     # Max number of job retries (for 500 errors) for the same job
@@ -244,7 +244,7 @@ worker:
     # Directory of the "datasets" library cache (for the datasets, not the modules)
     cacheDirectory: "/cache"
     # Git reference for the canonical datasets on https://github.com/huggingface/datasets
-    datasetsRevision: "master"
+    datasetsRevision: "main"
     # Log level
     logLevel: "INFO"
     # Max number of job retries (for 500 errors) for the same job

diff --git a/services/worker/.env.example b/services/worker/.env.example
@@ -5,7 +5,7 @@
 # ASSETS_DIRECTORY=
 
 # Git reference for the canonical datasets on https://github.com/huggingface/datasets
-# DATASETS_REVISION="master"
+# DATASETS_REVISION="main"
 
 # User Access Token (see https://huggingface.co/settings/token, only the `read` role is required)
 # HF_TOKEN=

diff --git a/services/worker/README.md b/services/worker/README.md
@@ -36,7 +36,7 @@ Set environment variables to configure the following aspects:
 
 - `ASSETS_BASE_URL`: base URL for the assets files. It should be set accordingly to the datasets-server domain, eg https://datasets-server.huggingface.co/assets. Defaults to `assets`.
 - `ASSETS_DIRECTORY`: directory where the asset files are stored. Defaults to empty, in which case the assets are located in the `datasets_server_assets` subdirectory inside the OS default cache directory.
-- `DATASETS_REVISION`: git reference for the canonical datasets on https://github.com/huggingface/datasets. Defaults to `master`.
+- `DATASETS_REVISION`: git reference for the canonical datasets on https://github.com/huggingface/datasets. Defaults to `main`.
 - `HF_DATASETS_CACHE`: directory where the `datasets` library will store the cached datasets data. Defaults to `~/.cache/huggingface/datasets`.
 - `HF_MODULES_CACHE`: directory where the `datasets` library will store the cached datasets scripts. Defaults to `~/.cache/huggingface/modules`.
 - `HF_TOKEN`: App Access Token (ask moonlanding administrators to get one, only the `read` role is required), to access the gated datasets. Defaults to empty.

diff --git a/services/worker/src/worker/constants.py b/services/worker/src/worker/constants.py
@@ -2,7 +2,7 @@
 
 DEFAULT_ASSETS_BASE_URL: str = "assets"
 DEFAULT_ASSETS_DIRECTORY: None = None
-DEFAULT_DATASETS_REVISION: str = "master"
+DEFAULT_DATASETS_REVISION: str = "main"
 DEFAULT_HF_TOKEN: Optional[str] = None
 DEFAULT_LOG_LEVEL: str = "INFO"
 DEFAULT_MAX_JOB_RETRIES: int = 3