Push to HF space

README.md

@@ -11,9 +11,9 @@ datasets:
   - lilacai/lilac-glaive-function-calling-v2
   - lilacai/lilac-open-assistant-conversations-2
   - lilacai/lilac-lmsys-chat-1m
-  - lilacai/lilac-OpenOrca
-  - lilacai/lilac-MMLU
   - lilacai/lilac-databricks-dolly-15k-curated-en
+  - lilacai/lilac-OpenOrca
   - lilacai/lilac-dolphin
+  - lilacai/lilac-mosaic-instruct-v3
 
 ---

data/lilac.yml

@@ -98,37 +98,6 @@ datasets:
             - content
       tags:
         - logs
-  - namespace: lilac
-    name: OpenOrca
-    source:
-      dataset_name: Open-Orca/OpenOrca
-      source_name: huggingface
-    embeddings:
-      - path: question
-        embedding: gte-small
-    settings:
-      ui:
-        media_paths:
-          - question
-          - response
-      tags:
-        - datasets
-  - namespace: lilac
-    name: MMLU
-    source:
-      dataset_name: cais/mmlu
-      config_name: all
-      source_name: huggingface
-    settings:
-      ui:
-        media_paths:
-          - question
-          - - choices
-            - '*'
-          - answer
-        markdown_paths: []
-      tags:
-        - eval
   - namespace: lilac
     name: databricks-dolly-15k-curated-en
     source:
@@ -158,6 +127,21 @@ datasets:
             - '*'
       tags:
         - machine-learning
+  - namespace: lilac
+    name: OpenOrca
+    source:
+      dataset_name: Open-Orca/OpenOrca
+      source_name: huggingface
+    embeddings:
+      - path: question
+        embedding: gte-small
+    settings:
+      ui:
+        media_paths:
+          - question
+          - response
+      tags:
+        - datasets
   - namespace: lilac
     name: dolphin
     tags:
@@ -175,6 +159,19 @@ datasets:
           - - input__cluster
             - text
         markdown_paths: []
+  - namespace: lilac
+    name: mosaic-instruct-v3
+    tags:
+      - datasets
+    source:
+      dataset_name: mosaicml/instruct-v3
+      source_name: huggingface
+    settings:
+      ui:
+        media_paths:
+          - prompt
+          - response
+        markdown_paths: []
 use_garden: true
 signals:
   - signal_name: text_statistics
@@ -222,11 +219,11 @@ clusters:
     dataset_name: databricks-dolly-15k-curated-en
     input_path: !!python/tuple
       - original-instruction
-  - dataset_namespace: lilac
-    dataset_name: MMLU
-    input_path: !!python/tuple
-      - question
   - dataset_namespace: lilac
     dataset_name: dolphin
     input_path: !!python/tuple
       - input
+  - dataset_namespace: lilac
+    dataset_name: mosaic-instruct-v3
+    input_path: !!python/tuple
+      - prompt