Push to HF space

.gitattributes

@@ -93,3 +93,5 @@ data/.cache/lilac/local/Capybara/conversation/\*/input__cluster.jsonl filter=lfs
 data/.cache/lilac/hf_download/lilacai/hermes-clusters-local-OpenHermes-2.5/local/OpenHermes-2.5/prompt__cluster-00000-of-00001.parquet filter=lfs diff=lfs merge=lfs -text
 data/.cache/lilac/hf_download/lilacai/hermes-clusters-local-OpenHermes-2.5/local/OpenHermes-2.5/data-00000-of-00001.parquet filter=lfs diff=lfs merge=lfs -text
 data/.cache/lilac/hf_download/lilacai/hermes-clusters-local-OpenHermes-2.5/local/OpenHermes-2.5/prompt-00000-of-00001.parquet filter=lfs diff=lfs merge=lfs -text
+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/hf_download/lilacai/nikhil_staging-local-Capybara/dataset_config.yml

@@ -0,0 +1,23 @@
+namespace: local
+name: Capybara
+source:
+  dataset_name: LDJnr/Capybara
+  source_name: huggingface
+signals:
+  - path:
+      - conversation
+      - '*'
+      - input
+    signal:
+      signal_name: text_statistics
+settings:
+  ui:
+    media_paths:
+      - - conversation
+        - '*'
+        - input
+      - - conversation
+        - '*'
+        - output
+      - input
+    markdown_paths: []

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/conversation/input/text_statistics/signal_manifest.json

@@ -0,0 +1,73 @@
+{
+  "files": [
+    "data-00000-of-00001.parquet"
+  ],
+  "parquet_id": "conversation.input.text_statistics",
+  "data_schema": {
+    "fields": {
+      "conversation": {
+        "repeated_field": {
+          "fields": {
+            "input": {
+              "fields": {
+                "text_statistics": {
+                  "fields": {
+                    "num_characters": {
+                      "dtype": {
+                        "type": "int32"
+                      }
+                    },
+                    "readability": {
+                      "dtype": {
+                        "type": "float32"
+                      }
+                    },
+                    "log(type_token_ratio)": {
+                      "dtype": {
+                        "type": "float32"
+                      }
+                    },
+                    "frac_non_ascii": {
+                      "dtype": {
+                        "type": "float32"
+                      },
+                      "bins": [
+                        [
+                          "Low",
+                          null,
+                          0.15
+                        ],
+                        [
+                          "Medium",
+                          0.15,
+                          0.3
+                        ],
+                        [
+                          "High",
+                          0.3,
+                          null
+                        ]
+                      ]
+                    }
+                  },
+                  "signal": {
+                    "signal_name": "text_statistics"
+                  }
+                }
+              }
+            }
+          }
+        }
+      }
+    }
+  },
+  "signal": {
+    "signal_name": "text_statistics"
+  },
+  "enriched_path": [
+    "conversation",
+    "*",
+    "input"
+  ],
+  "py_version": "0.2.0"
+}

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/data-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d65eea22b7c3bed4fe945881d287ecbbaa58f37f23e51da75e31d6b51da27494
+size 22848802

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/input-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e58ef67136014e702822e3791b38fd7b89e5dcd060a592865e1a959f7b95f5be
+size 7299202

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/input.map_manifest.json

@@ -0,0 +1,21 @@
+{
+  "files": [
+    "input-00000-of-00001.parquet"
+  ],
+  "parquet_id": "map.input",
+  "data_schema": {
+    "fields": {
+      "input": {
+        "dtype": {
+          "type": "string"
+        },
+        "map": {
+          "fn_name": "join",
+          "fn_source": "def join(item):\n  return '\\n'.join([x['input']['__value__'] for x in item['conversation']])\n",
+          "date_created": "2024-01-09T14:53:44.410899"
+        }
+      }
+    }
+  },
+  "py_version": "0.2.0"
+}

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/input__cluster.map_manifest.json

@@ -0,0 +1,50 @@
+{
+  "files": ["input__cluster-00000-of-00001.parquet"],
+  "parquet_id": "map.input__cluster",
+  "data_schema": {
+    "fields": {
+      "input__cluster": {
+        "fields": {
+          "cluster_id": {
+            "dtype": {
+              "type": "int32"
+            },
+            "categorical": true
+          },
+          "cluster_membership_prob": {
+            "dtype": {
+              "type": "float32"
+            }
+          },
+          "cluster_title": {
+            "dtype": {
+              "type": "string"
+            }
+          },
+          "category_id": {
+            "dtype": {
+              "type": "int32"
+            },
+            "categorical": true
+          },
+          "category_membership_prob": {
+            "dtype": {
+              "type": "float32"
+            }
+          },
+          "category_title": {
+            "dtype": {
+              "type": "string"
+            }
+          }
+        },
+        "cluster": {
+          "min_cluster_size": 5,
+          "remote": true,
+          "input_path": ["input"]
+        }
+      }
+    }
+  },
+  "py_version": "0.2.0"
+}

data/.cache/lilac/hf_download/lilacai/nikhil_staging-local-Capybara/local/Capybara/manifest.json

@@ -0,0 +1,40 @@
+{
+  "files": [
+    "data-00000-of-00001.parquet"
+  ],
+  "data_schema": {
+    "fields": {
+      "source": {
+        "dtype": {
+          "type": "string"
+        }
+      },
+      "conversation": {
+        "repeated_field": {
+          "fields": {
+            "input": {
+              "dtype": {
+                "type": "string"
+              }
+            },
+            "output": {
+              "dtype": {
+                "type": "string"
+              }
+            }
+          }
+        }
+      },
+      "__hfsplit__": {
+        "dtype": {
+          "type": "string"
+        }
+      }
+    }
+  },
+  "source": {
+    "dataset_name": "LDJnr/Capybara",
+    "load_from_disk": false,
+    "source_name": "huggingface"
+  }
+}

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

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:500b27604324ad6bae259c773d44486872a1754c0b89e6be97fe1b81a3e635e6
-size 2477986
+oid sha256:9ec9c9a251903209d6c3377ba6cc361a0d0533f87ce7f82d01b2df69ea1fe230
+size 2478036