Skip to content

Commit 4f1f64e

Browse files
rbs333rbs333
authored
RAAE-326 (#32)
* update llama-notebook
1 parent 573a962 commit 4f1f64e

File tree

1 file changed

+235
-28
lines changed

1 file changed

+235
-28
lines changed

‎python-recipes/RAG/03_llamaindex.ipynb‎

Lines changed: 235 additions & 28 deletions
Original file line numberDiff line numberDiff line change
@@ -33,30 +33,15 @@
3333
},
3434
{
3535
"cell_type": "code",
36-
"execution_count": 2,
36+
"execution_count": null,
3737
"metadata": {
3838
"colab": {
3939
"base_uri": "https://localhost:8080/"
4040
},
4141
"id": "UQezgPCG1vml",
4242
"outputId": "97b9bc03-da1b-439a-c37b-be6fdb58ab21"
4343
},
44-
"outputs": [
45-
{
46-
"name": "stdout",
47-
"output_type": "stream",
48-
"text": [
49-
"Cloning into 'temp_repo'...\n",
50-
"remote: Enumerating objects: 138, done.\u001b[K\n",
51-
"remote: Counting objects: 100% (138/138), done.\u001b[K\n",
52-
"remote: Compressing objects: 100% (98/98), done.\u001b[K\n",
53-
"remote: Total 138 (delta 68), reused 91 (delta 35), pack-reused 0\u001b[K\n",
54-
"Receiving objects: 100% (138/138), 7.19 MiB | 4.45 MiB/s, done.\n",
55-
"Resolving deltas: 100% (68/68), done.\n",
56-
"mv: rename temp_repo/resources to ./resources: Directory not empty\n"
57-
]
58-
}
59-
],
44+
"outputs": [],
6045
"source": [
6146
"# NBVAL_SKIP\n",
6247
"!git clone https://github.com/redis-developer/redis-ai-resources.git temp_repo\n",
@@ -200,7 +185,7 @@
200185
"name": "stdout",
201186
"output_type": "stream",
202187
"text": [
203-
"Sample doc Doc ID: 67e07154-6ea0-4822-8957-ac1d212fc9ee\n",
188+
"Sample doc Doc ID: c013353e-dae7-4d17-befd-9e784c8acf79\n",
204189
"Text: UNITED STATES SECURITIES AND EXCHANGE COMMISSION Washington,\n",
205190
"D.C. 20549 FORM 10-K (Mark One) ☒ ANNUAL REPORT PURSUANT T O SECTION\n",
206191
"13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934 For the fiscal year\n",
@@ -245,13 +230,13 @@
245230
},
246231
{
247232
"cell_type": "code",
248-
"execution_count": 10,
233+
"execution_count": 4,
249234
"metadata": {},
250235
"outputs": [],
251236
"source": [
252237
"from llama_index.core import StorageContext\n",
253238
"\n",
254-
"vector_store = RedisVectorStore(redis_url=REDIS_URL, index_name=\"llama\", overwrite=True)\n",
239+
"vector_store = RedisVectorStore(redis_url=REDIS_URL, overwrite=True)\n",
255240
"\n",
256241
"storage_context = StorageContext.from_defaults(vector_store=vector_store)\n",
257242
"\n",
@@ -267,7 +252,7 @@
267252
},
268253
{
269254
"cell_type": "code",
270-
"execution_count": 11,
255+
"execution_count": 5,
271256
"metadata": {},
272257
"outputs": [],
273258
"source": [
@@ -285,14 +270,14 @@
285270
},
286271
{
287272
"cell_type": "code",
288-
"execution_count": 12,
273+
"execution_count": 6,
289274
"metadata": {},
290275
"outputs": [
291276
{
292277
"name": "stdout",
293278
"output_type": "stream",
294279
"text": [
295-
"Node ID: b561dd17-5545-4d3a-bc4f-18cb39c7c01e\n",
280+
"Node ID: d2e6cd9c-0716-49d8-8563-407a00d05445\n",
296281
"Text: Table of Contents FISCAL 2023 NIKE BRAND REVENUE HIGHLIGHTS The\n",
297282
"following tables present NIKE Brand revenues disaggregated by\n",
298283
"reportable operating segment, distribution channel and major product\n",
@@ -301,14 +286,14 @@
301286
"fiscal 2022 on...\n",
302287
"Score: 0.900\n",
303288
"\n",
304-
"Node ID: 0415f059-9258-426b-8b21-34b287b3c21b\n",
289+
"Node ID: 28542d3b-b345-4e9e-b675-f62361ec85d9\n",
305290
"Text: Table of Contents NORTH AMERICA (Dollars in millions) FISCAL\n",
306291
"2023FISCAL 2022 % CHANGE% CHANGE EXCLUDING CURRENCY CHANGESFISCAL 2021\n",
307292
"% CHANGE% CHANGE EXCLUDING CURRENCY CHANGES Revenues by: Footwear $\n",
308293
"14,897 $ 12,228 22 % 22 %$ 11,644 5 % 5 % Apparel 5,947 5,492 8 % 9 %\n",
309294
"5,028 9 % 9 % Equipment 764 633 21 % 21 % 507 25 % 25 % TOTAL REVENUES\n",
310295
"$ 21,6...\n",
311-
"Score: 0.886\n",
296+
"Score: 0.885\n",
312297
"\n"
313298
]
314299
}
@@ -329,7 +314,7 @@
329314
},
330315
{
331316
"cell_type": "code",
332-
"execution_count": 13,
317+
"execution_count": 7,
333318
"metadata": {},
334319
"outputs": [
335320
{
@@ -338,7 +323,7 @@
338323
"\"NIKE's revenue in fiscal 23 was $51.2 billion.\""
339324
]
340325
},
341-
"execution_count": 13,
326+
"execution_count": 7,
342327
"metadata": {},
343328
"output_type": "execute_result"
344329
}
@@ -348,6 +333,228 @@
348333
"response.response"
349334
]
350335
},
336+
{
337+
"cell_type": "markdown",
338+
"metadata": {},
339+
"source": [
340+
"### Use a custom index schema\n",
341+
"\n",
342+
"In most use cases, you need the ability to customize the underling index configuration\n",
343+
"and specification. For example, this is handy in order to define specific metadata filters you wish to enable.\n",
344+
"\n",
345+
"With Redis, this is as simple as defining an index schema object\n",
346+
"(from file or dict) and passing it through to the vector store client wrapper."
347+
]
348+
},
349+
{
350+
"cell_type": "code",
351+
"execution_count": 8,
352+
"metadata": {},
353+
"outputs": [],
354+
"source": [
355+
"from redisvl.schema import IndexSchema\n",
356+
"\n",
357+
"\n",
358+
"custom_schema = IndexSchema.from_dict(\n",
359+
" {\n",
360+
" # customize basic index specs\n",
361+
" \"index\": {\n",
362+
" \"name\": \"custom_index\",\n",
363+
" \"prefix\": \"docs\",\n",
364+
" \"key_separator\": \":\",\n",
365+
" },\n",
366+
" # customize fields that are indexed\n",
367+
" \"fields\": [\n",
368+
" # required fields for llamaindex\n",
369+
" {\"type\": \"tag\", \"name\": \"id\"},\n",
370+
" {\"type\": \"tag\", \"name\": \"doc_id\"},\n",
371+
" {\"type\": \"text\", \"name\": \"text\"},\n",
372+
" # custom metadata fields\n",
373+
" {\"type\": \"numeric\", \"name\": \"updated_at\"},\n",
374+
" {\"type\": \"tag\", \"name\": \"file_name\"},\n",
375+
" # custom vector field definition for cohere embeddings\n",
376+
" {\n",
377+
" \"type\": \"vector\",\n",
378+
" \"name\": \"vector\",\n",
379+
" \"attrs\": {\n",
380+
" \"dims\": 1536,\n",
381+
" \"algorithm\": \"hnsw\",\n",
382+
" \"distance_metric\": \"cosine\",\n",
383+
" },\n",
384+
" },\n",
385+
" ],\n",
386+
" }\n",
387+
")"
388+
]
389+
},
390+
{
391+
"cell_type": "code",
392+
"execution_count": 9,
393+
"metadata": {},
394+
"outputs": [
395+
{
396+
"data": {
397+
"text/plain": [
398+
"IndexInfo(name='custom_index', prefix='docs', key_separator=':', storage_type=<StorageType.HASH: 'hash'>)"
399+
]
400+
},
401+
"execution_count": 9,
402+
"metadata": {},
403+
"output_type": "execute_result"
404+
}
405+
],
406+
"source": [
407+
"custom_schema.index"
408+
]
409+
},
410+
{
411+
"cell_type": "code",
412+
"execution_count": 10,
413+
"metadata": {},
414+
"outputs": [
415+
{
416+
"data": {
417+
"text/plain": [
418+
"{'id': TagField(name='id', type='tag', path=None, attrs=TagFieldAttributes(sortable=False, separator=',', case_sensitive=False, withsuffixtrie=False)),\n",
419+
" 'doc_id': TagField(name='doc_id', type='tag', path=None, attrs=TagFieldAttributes(sortable=False, separator=',', case_sensitive=False, withsuffixtrie=False)),\n",
420+
" 'text': TextField(name='text', type='text', path=None, attrs=TextFieldAttributes(sortable=False, weight=1, no_stem=False, withsuffixtrie=False, phonetic_matcher=None)),\n",
421+
" 'updated_at': NumericField(name='updated_at', type='numeric', path=None, attrs=NumericFieldAttributes(sortable=False)),\n",
422+
" 'file_name': TagField(name='file_name', type='tag', path=None, attrs=TagFieldAttributes(sortable=False, separator=',', case_sensitive=False, withsuffixtrie=False)),\n",
423+
" 'vector': HNSWVectorField(name='vector', type='vector', path=None, attrs=HNSWVectorFieldAttributes(dims=1536, algorithm=<VectorIndexAlgorithm.HNSW: 'HNSW'>, datatype=<VectorDataType.FLOAT32: 'FLOAT32'>, distance_metric=<VectorDistanceMetric.COSINE: 'COSINE'>, initial_cap=None, m=16, ef_construction=200, ef_runtime=10, epsilon=0.01))}"
424+
]
425+
},
426+
"execution_count": 10,
427+
"metadata": {},
428+
"output_type": "execute_result"
429+
}
430+
],
431+
"source": [
432+
"custom_schema.fields"
433+
]
434+
},
435+
{
436+
"cell_type": "code",
437+
"execution_count": 11,
438+
"metadata": {},
439+
"outputs": [],
440+
"source": [
441+
"# from datetime import datetime\n",
442+
"\n",
443+
"\n",
444+
"# def date_to_timestamp(date_string: str) -> int:\n",
445+
"# date_format: str = \"%Y-%m-%d\"\n",
446+
"# return int(datetime.strptime(date_string, date_format).timestamp())\n",
447+
"\n",
448+
"\n",
449+
"# # iterate through documents and add new field\n",
450+
"# for document in docs:\n",
451+
"# document.metadata[\"updated_at\"] = date_to_timestamp(\n",
452+
"# document.metadata[\"last_modified_date\"]\n",
453+
"# )"
454+
]
455+
},
456+
{
457+
"cell_type": "code",
458+
"execution_count": 12,
459+
"metadata": {},
460+
"outputs": [],
461+
"source": [
462+
"vector_store = RedisVectorStore(\n",
463+
" schema=custom_schema, # provide customized schema\n",
464+
" redis_url=REDIS_URL,\n",
465+
" overwrite=True,\n",
466+
")\n",
467+
"\n",
468+
"storage_context = StorageContext.from_defaults(vector_store=vector_store)\n",
469+
"\n",
470+
"# build and load index from documents and storage context\n",
471+
"index = VectorStoreIndex.from_documents(\n",
472+
" docs, storage_context=storage_context\n",
473+
")"
474+
]
475+
},
476+
{
477+
"cell_type": "markdown",
478+
"metadata": {},
479+
"source": [
480+
"### Query the vector store and filter on metadata\n",
481+
"Now that we have additional metadata indexed in Redis, let's try some queries which add in filters. As an example, we'll do a search for chunks with the word \"audit\" from an exact file \"amzn-10k-2023.pdf\". "
482+
]
483+
},
484+
{
485+
"cell_type": "code",
486+
"execution_count": 13,
487+
"metadata": {},
488+
"outputs": [],
489+
"source": [
490+
"from llama_index.core.vector_stores import (\n",
491+
" MetadataFilters,\n",
492+
" MetadataFilter,\n",
493+
" ExactMatchFilter,\n",
494+
")\n",
495+
"\n",
496+
"retriever = index.as_retriever(\n",
497+
" similarity_top_k=3,\n",
498+
" filters=MetadataFilters(\n",
499+
" filters=[\n",
500+
" ExactMatchFilter(key=\"file_name\", value=\"amzn-10k-2023.pdf\"),\n",
501+
" MetadataFilter(\n",
502+
" key=\"text\",\n",
503+
" value=\"audit\",\n",
504+
" operator=\"text_match\",\n",
505+
" ),\n",
506+
" ],\n",
507+
" condition=\"and\",\n",
508+
" ),\n",
509+
")"
510+
]
511+
},
512+
{
513+
"cell_type": "code",
514+
"execution_count": 14,
515+
"metadata": {},
516+
"outputs": [
517+
{
518+
"name": "stdout",
519+
"output_type": "stream",
520+
"text": [
521+
"Node ID: cd0c5d8f-e3b1-4cbb-aa6a-5960003cdb2d\n",
522+
"Text: Table of Contents valuation. In the ordinary course of our\n",
523+
"business, there are many transactions and calculations for which the\n",
524+
"ultimate tax determination is uncertain. Significant judgment is\n",
525+
"required in evaluating and estimating our tax expense, assets, and\n",
526+
"liabilities. We are also subject to tax controversies in various\n",
527+
"jurisdictions that can...\n",
528+
"Score: 0.746\n",
529+
"\n",
530+
"Node ID: 6745f668-4c7a-43bf-a9c3-9b04e1a497f8\n",
531+
"Text: Table of Contents Included in other income (expense), net in\n",
532+
"2021 and 2022 is a marketable equity securities valuation gain (loss)\n",
533+
"of $11.8 billion and $(12.7) billion from our equity investment in\n",
534+
"Rivian Automotive, Inc. (“Rivian”). Our investment in Rivian’s\n",
535+
"preferred stock was accounted for at cost, with adjustments for\n",
536+
"observable changes in ...\n",
537+
"Score: 0.740\n",
538+
"\n",
539+
"Node ID: 717666fe-fea5-488b-999c-84e6d8b9a0db\n",
540+
"Text: Exhibit 31.1 CERTIFICATIONS I, Andrew R. Jassy, certify that: 1.\n",
541+
"I have reviewed this Form 10-K of Amazon.com, Inc.; 2. Based on my\n",
542+
"knowledge, this report does not contain any untrue statement of a\n",
543+
"material fact or omit to state a material fact necessary to make the\n",
544+
"statements made, in light of the circumstances under which such\n",
545+
"statements were ...\n",
546+
"Score: 0.732\n",
547+
"\n"
548+
]
549+
}
550+
],
551+
"source": [
552+
"result_nodes = retriever.retrieve(\"What did the author learn?\")\n",
553+
"\n",
554+
"for node in result_nodes:\n",
555+
" print(node)"
556+
]
557+
},
351558
{
352559
"cell_type": "code",
353560
"execution_count": null,
@@ -376,7 +583,7 @@
376583
"name": "python",
377584
"nbconvert_exporter": "python",
378585
"pygments_lexer": "ipython3",
379-
"version": "3.9.12"
586+
"version": "3.11.9"
380587
},
381588
"widgets": {
382589
"application/vnd.jupyter.widget-state+json": {

0 commit comments

Comments
 (0)