Skip to content

armgit / txtai

Public Imported from GitHub
0

💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows

Update notebook

davidmezzettib43cdc1f4354
diff --git a/examples/24_Whats_new_in_txtai_4_0.ipynb b/examples/24_Whats_new_in_txtai_4_0.ipynb
index 424aef69..13de0b14 100644
--- a/examples/24_Whats_new_in_txtai_4_0.ipynb
+++ b/examples/24_Whats_new_in_txtai_4_0.ipynb
@@ -63,7 +63,7 @@
"%%capture\n",
"!pip install git+https://github.com/neuml/txtai"
],
- "execution_count": null,
+ "execution_count": 21,
"outputs": []
},
{
@@ -90,7 +90,7 @@
"colab": {
"base_uri": "https://localhost:8080/"
},
- "outputId": "8c7ff0df-e447-47fc-a5b5-480fcd4df01d"
+ "outputId": "c37d3b1d-1096-4518-92fa-1950cb69b183"
},
"source": [
"from txtai.embeddings import Embeddings\n",
@@ -119,7 +119,7 @@
" # Print text\n",
" print(\"%-20s %s\" % (query, text))"
],
- "execution_count": null,
+ "execution_count": 22,
"outputs": [
{
"output_type": "stream",
@@ -182,9 +182,9 @@
"base_uri": "https://localhost:8080/"
},
"id": "aPH-dnV2ZuL1",
- "outputId": "13b58387-0a76-466e-84bc-b51c8f0427dd"
+ "outputId": "7946a782-8b9c-4ee5-abcc-ce0f45858349"
},
- "execution_count": null,
+ "execution_count": 23,
"outputs": [
{
"output_type": "stream",
@@ -251,9 +251,9 @@
"base_uri": "https://localhost:8080/",
"height": 420
},
- "outputId": "2eeec5fc-260b-42f0-eae5-4957da37684c"
+ "outputId": "3dfe3593-6a07-4277-b388-4983e81c5e8a"
},
- "execution_count": null,
+ "execution_count": 24,
"outputs": [
{
"output_type": "execute_result",
@@ -268,7 +268,7 @@
"width": 600
}
},
- "execution_count": 40
+ "execution_count": 24
}
]
},
@@ -302,9 +302,9 @@
"base_uri": "https://localhost:8080/"
},
"id": "fAz7Jv6SjR6a",
- "outputId": "d9b6f387-3c5b-4fd2-968d-bcee43d91ae2"
+ "outputId": "0ea984bf-a9ab-42fc-fa18-3f59bd8ef2b9"
},
- "execution_count": null,
+ "execution_count": 25,
"outputs": [
{
"output_type": "stream",
@@ -313,7 +313,7 @@
"{\n",
" \"backend\": \"faiss\",\n",
" \"build\": {\n",
- " \"create\": \"2021-12-28T14:29:24Z\",\n",
+ " \"create\": \"2021-12-29T00:42:17Z\",\n",
" \"python\": \"3.7.12\",\n",
" \"settings\": {\n",
" \"components\": \"IDMap,Flat\"\n",
@@ -331,7 +331,7 @@
"{\n",
" \"backend\": \"faiss\",\n",
" \"build\": {\n",
- " \"create\": \"2021-12-28T14:29:30Z\",\n",
+ " \"create\": \"2021-12-29T00:42:29Z\",\n",
" \"python\": \"3.7.12\",\n",
" \"settings\": {\n",
" \"components\": \"IDMap,Flat\"\n",
@@ -383,18 +383,18 @@
"base_uri": "https://localhost:8080/"
},
"id": "0oOC8ToG1pyn",
- "outputId": "bba1aa74-ef50-4137-bc5f-38f1961dd794"
+ "outputId": "bd42dbbe-0de1-4fe3-97b9-602992bdca5a"
},
- "execution_count": null,
+ "execution_count": 26,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
- "drwx------ root/root 0 2021-12-28 14:29 ./\n",
- "-rw-r--r-- root/root 290 2021-12-28 14:29 ./config\n",
- "-rw-r--r-- root/root 77824 2021-12-28 14:29 ./documents\n",
- "-rw-r--r-- root/root 10898 2021-12-28 14:29 ./embeddings\n",
+ "drwx------ root/root 0 2021-12-29 00:42 ./\n",
+ "-rw-r--r-- root/root 290 2021-12-29 00:42 ./config\n",
+ "-rw-r--r-- root/root 77824 2021-12-29 00:42 ./documents\n",
+ "-rw-r--r-- root/root 10898 2021-12-29 00:42 ./embeddings\n",
"\n",
"Strms Blocks Compressed Uncompressed Ratio Check Filename\n",
" 1 1 45.8 KiB 100.0 KiB 0.458 CRC64 index.tar.xz\n",
@@ -411,7 +411,7 @@
]
},
"metadata": {},
- "execution_count": 42
+ "execution_count": 26
}
]
},
@@ -424,6 +424,68 @@
"id": "x5iSRKZVYfok"
}
},
+ {
+ "cell_type": "markdown",
+ "source": [
+ "# External vector models\n",
+ "\n",
+ "txtai supports generating vectors with [Hugging Face Transformers](https://github.com/huggingface/transformers), [PyTorch](https://github.com/pytorch/pytorch), [ONNX](https://github.com/microsoft/onnxruntime) and [Word Vector](https://github.com/neuml/magnitude) models.\n",
+ "\n",
+ "This release adds support for pre-computed vectors using external models. External models may be an API, custom library and/or another way to vectorize data. The example below uses [spaCy](https://github.com/explosion/spaCy) to build vectors and then loads them into txtai. "
+ ],
+ "metadata": {
+ "id": "oAISaBYVg60i"
+ }
+ },
+ {
+ "cell_type": "code",
+ "source": [
+ "%%capture\n",
+ "!pip install spacy --upgrade\n",
+ "!python -m spacy download en_core_web_md"
+ ],
+ "metadata": {
+ "id": "IIGlel3ShyGP"
+ },
+ "execution_count": 27,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "source": [
+ "import spacy\n",
+ "\n",
+ "# Load spacy\n",
+ "nlp = spacy.load(\"en_core_web_md\")\n",
+ "\n",
+ "def transform(document):\n",
+ " return nlp(document[1]).vector\n",
+ "\n",
+ "# Index data with spacy pipeline\n",
+ "embeddings = Embeddings({\"method\": \"external\", \"transform\": transform, \"content\": True})\n",
+ "embeddings.index([(x, {\"object\": nlp(row).vector, \"row\": row}, None) for x, row in enumerate(data)])\n",
+ "\n",
+ "# Run search\n",
+ "print(embeddings.search(\"select id, row as text, score from txtai where similar('hiking danger')\", 1))"
+ ],
+ "metadata": {
+ "colab": {
+ "base_uri": "https://localhost:8080/"
+ },
+ "id": "nikKybxEkhjC",
+ "outputId": "84950459-8d2b-4570-b7ed-ad468c783901"
+ },
+ "execution_count": 28,
+ "outputs": [
+ {
+ "output_type": "stream",
+ "name": "stdout",
+ "text": [
+ "[{'id': '3', 'text': 'The National Park Service warns against sacrificing slower friends in a bear attack', 'score': 0.5822835564613342}]\n"
+ ]
+ }
+ ]
+ },
{
"cell_type": "markdown",
"metadata": {