Push

.gitattributes

@@ -1,3 +0,0 @@
-dist/lilac-0.0.17-py3-none-any.whl filter=lfs diff=lfs merge=lfs -text
-data/.cache/lilac/concept/lilac/profanity/gte-small.pkl filter=lfs diff=lfs merge=lfs -text
-data/.cache/lilac/concept/lilac/toxicity/gte-small.pkl filter=lfs diff=lfs merge=lfs -text

data/.cache/lilac/concept/lilac/profanity/gte-small.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ed7340614b1dea910ddeb26bbda0167b1f4fe2479071a62a70b63c18bc6232d0
-size 1672960

data/.cache/lilac/concept/lilac/toxicity/gte-small.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f2af2736f3d749391a431f9c24d3fc78cf8e58457cc4f0d1ce770185b92d879c
-size 1886446

data/lilac.yml

@@ -0,0 +1,226 @@
+# Lilac project config.
+# See https://lilacml.com/api_reference/index.html#lilac.Config for details.
+
+datasets:
+  - namespace: local
+    name: glue
+    source:
+      dataset_name: glue
+      config_name: ax
+      source_name: huggingface
+    embeddings:
+      - path: premise
+        embedding: gte-small
+    signals:
+      - path: premise
+        signal:
+          signal_name: pii
+      - path: hypothesis
+        signal:
+          signal_name: pii
+    settings:
+      ui:
+        media_paths:
+          - premise
+  - namespace: local
+    name: glue_ax
+    source:
+      dataset_name: glue
+      config_name: ax
+      source_name: huggingface
+    embeddings:
+      - path: hypothesis
+        embedding: gte-small
+    settings:
+      ui:
+        media_paths:
+          - hypothesis
+  - namespace: local
+    name: imdb3
+    source:
+      dataset_name: imdb
+      source_name: huggingface
+    settings:
+      ui:
+        media_paths:
+          - text
+  - namespace: local
+    name: imdb
+    source:
+      dataset_name: imdb
+      source_name: huggingface
+    embeddings:
+      - path: text
+        embedding: gte-small
+    settings:
+      ui:
+        media_paths:
+          - text
+  - namespace: local
+    name: imdb2
+    source:
+      dataset_name: imdb
+      source_name: huggingface
+    settings:
+      ui:
+        media_paths:
+          - text
+  - namespace: lilac
+    name: OpenOrca-100k
+    source:
+      dataset_name: Open-Orca/OpenOrca
+      sample_size: 100000
+      source_name: huggingface
+    embeddings:
+      - path: question
+        embedding: gte-small
+      - path: response
+        embedding: gte-small
+    signals:
+      - path: question
+        signal:
+          signal_name: near_dup
+      - path: question
+        signal:
+          signal_name: pii
+      - path: question
+        signal:
+          signal_name: lang_detection
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          signal_name: concept_score
+      - path: question
+        signal:
+          signal_name: text_statistics
+      - path: response
+        signal:
+          signal_name: near_dup
+      - path: response
+        signal:
+          signal_name: pii
+      - path: response
+        signal:
+          signal_name: lang_detection
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          signal_name: concept_score
+      - path: response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          signal_name: concept_score
+      - path: response
+        signal:
+          signal_name: text_statistics
+    settings:
+      ui:
+        media_paths:
+          - question
+          - response
+      preferred_embedding: gte-small
+  - namespace: local
+    name: the_movies_dataset
+    source:
+      filepaths:
+        - gs://lilac-data/datasets/the_movies_dataset/the_movies_dataset.csv
+      names: []
+      source_name: csv
+    settings:
+      ui:
+        media_paths:
+          - overview
+  - namespace: local
+    name: glue_ax_parquet
+    source:
+      filepaths:
+        - gs://lilac-data/datasets/glue_ax_parquet/glue_ax.parquet
+      source_name: parquet
+    settings:
+      ui:
+        media_paths:
+          - premise

dist/README.md

@@ -1,2 +0,0 @@
-This directory is used for locally built whl files.
-We write a README.md to ensure an empty folder is uploaded when there is no whl.

dist/lilac-0.0.17-py3-none-any.whl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:0af64bad5f5bfea7c8bddef1804297b3e518c82742c1edb18f50155ad1e4f69c
-size 1132299

docker_start.py

@@ -9,6 +9,7 @@ from huggingface_hub import scan_cache_dir, snapshot_download
 
 from lilac.concepts.db_concept import CONCEPTS_DIR, DiskConceptDB, get_concept_output_dir
 from lilac.env import data_path, env
+from lilac.project import PROJECT_CONFIG_FILENAME
 from lilac.utils import get_datasets_dir, get_lilac_cache_dir, log
 
 
@@ -70,8 +71,12 @@ def main() -> None:
       ignore_patterns=['.gitattributes', 'README.md'])
 
   snapshot_dir = snapshot_download(repo_id=repo_id, repo_type='space', token=env('HF_ACCESS_TOKEN'))
-  # Copy datasets.
+
   spaces_data_dir = os.path.join(snapshot_dir, 'data')
+  # Copy the config file.
+  project_config_file = os.path.join(spaces_data_dir, PROJECT_CONFIG_FILENAME)
+  if os.path.exists(project_config_file):
+    shutil.copy(project_config_file, os.path.join(data_path(), PROJECT_CONFIG_FILENAME))
 
   # Delete cache files from persistent storage.
   cache_dir = get_lilac_cache_dir(data_path())