From 363f159e5ab893cd8a9400b1784cfeea3654351e Mon Sep 17 00:00:00 2001
From: Nikhil Thorat <nsthorat@gmail.com>
Date: Fri, 26 Jan 2024 11:32:57 -0500
Subject: [PATCH 1/3] Add hacker news comments to the public demo.

---
 lilac_hf_space.yml | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/lilac_hf_space.yml b/lilac_hf_space.yml
index 90b88212..5366d44f 100644
--- a/lilac_hf_space.yml
+++ b/lilac_hf_space.yml
@@ -173,6 +173,23 @@ datasets:
         path:
           - completion
 
+  - namespace: lilac
+    name: hncomments-1m
+    source:
+      filepaths:
+        - /Users/brian/dev/lilac/data/datasets/local/hncomments-duckprogress/data-00000-of-00001.parquet
+      sample_size: 1000000
+      source_name: parquet
+    settings:
+      tags: [datasets]
+      ui:
+        media_paths:
+          - text
+    embeddings:
+      - embedding: gte-small
+        path:
+          - text
+
   ## Eval datasets
   - namespace: lilac
     name: MMLU

From caeaa4637fe6cd1e58d44a4d2a382a76db1d86e1 Mon Sep 17 00:00:00 2001
From: Nikhil Thorat <nsthorat@gmail.com>
Date: Fri, 26 Jan 2024 11:33:30 -0500
Subject: [PATCH 2/3] save

---
 lilac_hf_space.yml | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/lilac_hf_space.yml b/lilac_hf_space.yml
index 5366d44f..31860533 100644
--- a/lilac_hf_space.yml
+++ b/lilac_hf_space.yml
@@ -497,6 +497,11 @@ clusters:
     output_path:
       - messages__clusters
 
+  - dataset_namespace: lilac
+    dataset_name: hncomments-1m
+    input_path:
+      - text
+
   # NOTE: We cluster both prompt and completion because the roblox prompts are just small sections
   # of code which are not extremely descriptive.
   - dataset_namespace: lilac

From e69d5015a7d9273e335b8ee9f4df9c2fc2c21af4 Mon Sep 17 00:00:00 2001
From: Nikhil Thorat <nsthorat@gmail.com>
Date: Fri, 26 Jan 2024 12:01:13 -0500
Subject: [PATCH 3/3] save

---
 lilac_hf_space.yml | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/lilac_hf_space.yml b/lilac_hf_space.yml
index 31860533..c09d594c 100644
--- a/lilac_hf_space.yml
+++ b/lilac_hf_space.yml
@@ -176,10 +176,9 @@ datasets:
   - namespace: lilac
     name: hncomments-1m
     source:
-      filepaths:
-        - /Users/brian/dev/lilac/data/datasets/local/hncomments-duckprogress/data-00000-of-00001.parquet
+      dataset_name: OpenPipe/hacker-news
       sample_size: 1000000
-      source_name: parquet
+      source_name: huggingface
     settings:
       tags: [datasets]
       ui: