Push to HF space

README.md

@@ -9,6 +9,7 @@ datasets:
   - lilacai/lilac-Capybara
   - lilacai/lilac-glaive-code-assistant
   - lilacai/lilac-open-assistant-conversations-2
+  - lilacai/lilac-lmsys-chat-1m
   - lilacai/lilac-databricks-dolly-15k-curated-en
   - lilacai/lilac-OpenOrca-100k
   - lilacai/lilac-dolphin

data/lilac.yml

@@ -59,6 +59,25 @@ datasets:
           - text
       tags:
         - datasets
+  - namespace: lilac
+    name: lmsys-chat-1m
+    source:
+      dataset_name: OpenAssistant/oasst2
+      source_name: huggingface
+    embeddings:
+      - path:
+          - conversation
+          - '*'
+          - content
+        embedding: gte-small
+    settings:
+      ui:
+        media_paths:
+          - - conversation
+            - '*'
+            - content
+      tags:
+        - logs
   - namespace: lilac
     name: databricks-dolly-15k-curated-en
     source:
@@ -127,10 +146,6 @@ use_garden: true
 signals:
   - signal_name: text_statistics
   - signal_name: lang_detection
-  - embedding: gte-small
-    namespace: lilac
-    concept_name: profanity
-    signal_name: concept_score
 concept_model_cache_embeddings:
   - gte-small
   - gte-base
@@ -152,6 +167,13 @@ clusters:
     dataset_name: open-assistant-conversations-2
     input_path: !!python/tuple
       - text
+  - dataset_namespace: lilac
+    dataset_name: lmsys-chat-1m
+    input_selector:
+      format: openai_json
+      selector: user
+    output_path: !!python/tuple
+      - conversation__clusters
   - dataset_namespace: lilac
     dataset_name: databricks-dolly-15k-curated-en
     input_path: !!python/tuple

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74732a23985fa54c07ace5aafd0311c998c13d7a17e8cb25ae511054638b56ad
-size 2478879
+oid sha256:e51b13532baa6dc500a59207750d21bc56d99f566590ec85c6c07b91bce7ab25
+size 2480452