Push to HF space

.gitattributes

@@ -96,3 +96,4 @@ data/.cache/lilac/hf_download/lilacai/hermes-clusters-local-OpenHermes-2.5/local
 data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/data-00000-of-00001.parquet filter=lfs diff=lfs merge=lfs -text
 data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/input-00000-of-00001.parquet filter=lfs diff=lfs merge=lfs -text
 data/.cache/lilac/concept/lilac/question/nomic-embed-1.5-768.pkl filter=lfs diff=lfs merge=lfs -text
+data/.cache/lilac/concept/lilac/toxicity/openai.pkl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -6,6 +6,7 @@ colorTo: purple
 sdk: docker
 app_port: 5432
 datasets:
-  - lilacai/nikhil_staging-local-glue_ax_map
+  - lilacai/nikhil_staging-local-OpenHermes-2.5-10k
+  - lilacai/nikhil_staging-local-Capybara
 
 ---

data/.cache/lilac/concept/lilac/toxicity/openai.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:722c5141fb02f101c45b0530ae2de38685716b03b8bd52592a41f6e1c62dafc9
+size 7384504

data/lilac.yml

@@ -1,34 +1,53 @@
 datasets:
   - namespace: local
-    name: glue_ax_map
+    name: OpenHermes-2.5-10k
     source:
-      dataset_name: glue
-      config_name: ax
+      dataset_name: teknium/OpenHermes-2.5
+      sample_size: 9999
       source_name: huggingface
     embeddings:
-      - path: premise
-        embedding: bge-m3
-      - path: premise
-        embedding: nomic-embed-1.5-768
-      - path: hypothesis
-        embedding: nomic-embed-1.5-768
+      - path:
+          - conversations
+          - '*'
+          - value
+        embedding: gte-small
+    settings:
+      ui:
+        media_paths:
+          - - test__clusters
+            - text
+          - - conversations
+            - '*'
+            - value
+          - - test__cluster
+            - text
+        markdown_paths: []
+  - namespace: local
+    name: Capybara
+    source:
+      dataset_name: LDJnr/Capybara
+      source_name: huggingface
+    embeddings:
+      - path:
+          - conversation
+          - '*'
+          - input
+        embedding: gte-small
     signals:
-      - path: premise
+      - path:
+          - conversation
+          - '*'
+          - input
         signal:
-          signal_name: test_signal
-      - path: premise
-        signal:
-          signal_name: test_signal
+          signal_name: text_statistics
     settings:
       ui:
         media_paths:
-          - premise
-          - hypothesis
-          - premise_upper
-          - premise_upper2
-          - premise_flat
-          - - hypothesis_upper
-            - value
-          - - hypothesis_upper2
-            - value
+          - input
+          - - conversation
+            - '*'
+            - input
+          - - conversation
+            - '*'
+            - output
         markdown_paths: []

dist/lilac-1337.0.0-py3-none-any.whl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5ba686997302c6a2adee46a95a702a99689c9ca589fd6150506113bc3652b07c
-size 2490799
+oid sha256:05af7f9fef18ea8c13248b3fab700de75f05787510e98d364b0090a67f640df3
+size 2494497