Repository navigation
Expand file tree
/
Copy pathllms.txt
More file actions
11411 lines (8531 loc) · 331 KB
/
Copy pathllms.txt
File metadata and controls
11411 lines (8531 loc) · 331 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Understanding Internals of R2R Library
## Table of Contents
1. [Introduction](#introduction)
2. [Installation](#installation)
- [Prerequisites](#prerequisites)
- [Docker Installation](#docker-installation)
- [Install the R2R CLI & Python SDK](#install-the-r2r-cli--python-sdk)
- [Start R2R with Docker](#start-r2r-with-docker)
- [Google Cloud Platform Deployment](#google-cloud-platform-deployment)
- [Overview](#overview)
- [Creating a Google Compute Engine Instance](#creating-a-google-compute-engine-instance)
- [Installing Dependencies](#installing-dependencies)
- [Setting up R2R](#setting-up-r2r)
- [Configuring Port Forwarding for Local Access](#configuring-port-forwarding-for-local-access)
- [Exposing Ports for Public Access (Optional)](#exposing-ports-for-public-access-optional)
- [Conclusion](#conclusion-1)
3. [R2R Application Lifecycle](#r2r-application-lifecycle)
- [Developer Workflow](#developer-workflow)
- [User Interaction](#user-interaction)
- [Hello R2R (Code Example)](#hello-r2r-code-example)
4. [Configuration](#configuration)
- [Configuration Overview](#configuration-overview)
- [Server-Side Configuration (`r2r.toml`)](#server-side-configuration-r2rtoml)
- [Example: `r2r.toml`](#example-r2rtoml)
- [Runtime Overrides](#runtime-overrides)
- [Postgres Configuration](#postgres-configuration)
- [Example Configuration](#example-configuration-1)
- [Key Features](#key-features)
- [Embedding Configuration](#embedding-configuration)
- [Example Configuration](#example-configuration-2)
- [Auth & Users Configuration](#auth--users-configuration)
- [Example Configuration](#example-configuration-3)
- [Key Features](#key-features-1)
- [Data Ingestion Configuration](#data-ingestion-configuration)
- [Example Configuration](#example-configuration-4)
- [Retrieval Configuration](#retrieval-configuration)
- [Example Configuration](#example-configuration-5)
- [RAG Configuration](#rag-configuration)
- [Example Configuration](#example-configuration-6)
- [Graphs Configuration](#graphs-configuration)
- [Example Configuration](#example-configuration-7)
- [Prompts Configuration](#prompts-configuration)
- [Example Configuration](#example-configuration-8)
5. [Data Ingestion](#data-ingestion)
- [Introduction](#introduction-1)
- [Ingestion Modes](#ingestion-modes)
- [Ingesting Documents](#ingesting-documents)
- [Example Response](#example-response)
- [Ingesting Pre-Processed Chunks](#ingesting-pre-processed-chunks)
- [Example](#example-1)
- [Deleting Documents and Chunks](#deleting-documents-and-chunks)
- [Delete a Document](#delete-a-document)
- [Sample Output](#sample-output)
- [Key Features of Deletion](#key-features-of-deletion)
- [Additional Configuration & Concepts](#additional-configuration--concepts)
- [Light vs. Full Deployments](#light-vs-full-deployments)
- [Provider Configuration](#provider-configuration)
- [Conclusion](#conclusion-2)
6. [Contextual Enrichment](#contextual-enrichment)
- [The Challenge of Context Loss](#the-challenge-of-context-loss)
- [Introducing Contextual Enrichment](#introducing-contextual-enrichment)
- [Enabling Enrichment](#enabling-enrichment)
- [Enrichment Strategies Explained](#enrichment-strategies-explained)
- [Neighborhood Strategy](#neighborhood-strategy)
- [Semantic Strategy](#semantic-strategy)
- [The Enrichment Process](#the-enrichment-process)
- [Implementation and Results](#implementation-and-results)
- [Viewing Enriched Results](#viewing-enriched-results)
- [Metadata and Storage](#metadata-and-storage)
- [Best Practices](#best-practices-1)
- [Conclusion](#conclusion-3)
7. [AI Powered Search](#ai-powered-search)
- [Introduction](#introduction-2)
- [Understanding Search Modes](#understanding-search-modes)
- [How R2R Hybrid Search Works](#how-r2r-hybrid-search-works)
- [Vector Search](#vector-search)
- [Example](#example-2)
- [Hybrid Search](#hybrid-search)
- [Example](#example-3)
- [Knowledge Graph Search](#knowledge-graph-search)
- [Example](#example-4)
- [Reciprocal Rank Fusion (RRF)](#reciprocal-rank-fusion-rrf)
- [Result Ranking](#result-ranking)
- [Configuration](#configuration-1)
- [Choosing a Search Mode](#choosing-a-search-mode)
- [Best Practices](#best-practices-2)
- [Conclusion](#conclusion-4)
8. [Retrieval-Augmented Generation (RAG)](#retrieval-augmented-generation-rag)
- [Basic RAG](#basic-rag)
- [Example](#example-5)
- [Sample Output](#sample-output-1)
- [RAG with Hybrid Search](#rag-w-hybrid-search)
- [Example](#example-6)
- [Streaming RAG](#streaming-rag)
- [Example](#example-7)
- [Customizing RAG](#customizing-rag)
- [Example](#example-8)
- [Advanced RAG Techniques](#advanced-rag-techniques)
- [HyDE (Hypothetical Document Embeddings)](#hyde-hypothetical-document-embeddings)
- [Workflow](#workflow)
- [Python Example](#python-example-1)
- [Sample Output](#sample-output-2)
- [RAG-Fusion](#rag-fusion)
- [Workflow](#workflow-1)
- [Python Example](#python-example-2)
- [Sample Output](#sample-output-3)
- [Combining with Other Settings](#combining-with-other-settings)
- [Example](#example-9)
- [Customization and Server-Side Defaults](#customization-and-server-side-defaults)
- [Example](#example-10)
- [Conclusion](#conclusion-5)
9. [Knowledge Graphs in R2R](#knowledge-graphs-in-r2r)
- [Overview](#overview-2)
- [System Architecture](#system-architecture)
- [Getting Started](#getting-started)
- [Document-Level Extraction](#document-level-extraction)
- [Python Example](#python-example-3)
- [Creating Collection Graphs](#creating-collection-graphs)
- [Python Example](#python-example-4)
- [Managing Collection Graphs](#managing-collection-graphs)
- [Python Example](#python-example-5)
- [Example Output](#example-output-4)
- [Graph-Collection Relationship](#graph-collection-relationship)
- [Knowledge Graph Workflow](#knowledge-graph-workflow)
- [Step 1: Extract Document Knowledge](#step-1-extract-document-knowledge)
- [Step 2: Initialize and Populate Graph](#step-2-initialize-and-populate-graph)
- [Step 3: View Entities and Relationships](#step-3-view-entities-and-relationships)
- [Step 4: Build Graph Communities](#step-4-build-graph-communities)
- [Step 5: KG-Enhanced Search](#step-5-kg-enhanced-search)
- [Step 6: Reset Graph](#step-6-reset-graph)
- [Graph Synchronization](#graph-synchronization)
- [Document Updates](#document-updates)
- [Cross-Collection Updates](#cross-collection-updates)
- [Access Control](#access-control)
- [Python Example](#python-example-6)
- [Using Knowledge Graphs](#using-knowledge-graphs)
- [Search Integration](#search-integration)
- [Curl Example](#curl-example-1)
- [RAG Integration](#rag-integration)
- [Python Example](#python-example-7)
- [Best Practices]
10BC8
(#best-practices-3)
- [Document Management](#document-management)
- [Collection Management](#collection-management)
- [Performance Optimization](#performance-optimization)
- [Access Control](#access-control-1)
- [Troubleshooting](#troubleshooting-1)
- [Conclusion](#conclusion-6)
- [Next Steps](#next-steps-1)
10. [GraphRAG in R2R](#graphrag-in-r2r)
- [Overview](#overview-1)
- [Architecture](#architecture)
- [Understanding Communities](#understanding-communities)
- [Example Communities](#example-communities)
- [Implementation Guide](#implementation-guide)
- [Prerequisites](#prerequisites-1)
- [Python Example](#python-example-8)
- [Building Communities](#building-communities)
- [Python Example](#python-example-9)
- [Build Process Includes](#build-process-includes)
- [Using GraphRAG](#using-graphrag)
- [Python Example](#python-example-10)
- [Understanding Results](#understanding-results)
- [Document Chunks](#document-chunks)
- [Graph Elements](#graph-elements)
- [Communities](#communities-1)
- [Scaling GraphRAG](#scaling-graphrag)
- [Using Orchestration](#using-orchestration)
- [Access Hatchet UI](#access-hatchet-ui)
- [Features](#features-1)
- [Example Diagram](#example-diagram)
- [Best Practices](#best-practices-4)
- [Development](#development)
- [Performance](#performance-1)
- [Quality](#quality)
- [Troubleshooting](#troubleshooting-2)
- [Next Steps](#next-steps-2)
- [Conclusion](#conclusion-7)
- [Security Considerations](#security-considerations-1)
11. [Agent](#agent)
- [Understanding R2R’s RAG Agent](#understanding-r2rs-rag-agent)
- [Planned Extensions](#planned-extensions)
- [Configuration](#configuration-2)
- [Default Configuration](#default-configuration)
- [Enable Web Search](#enable-web-search)
- [Using the RAG Agent](#using-the-rag-agent)
- [Python Example](#python-example-11)
- [Streaming Responses](#streaming-responses)
- [Context-Aware Responses](#context-aware-responses)
- [Working with Files](#working-with-files)
- [Python Example](#python-example-12)
- [Advanced Features](#advanced-features)
- [Combined Search Capabilities](#combined-search-capabilities)
- [Example](#example-11)
- [Custom Search Settings](#custom-search-settings)
- [Example](#example-12)
- [Best Practices](#best-practices-5)
- [Conversation Management](#conversation-management)
- [Search Optimization](#search-optimization)
- [Response Handling](#response-handling)
- [Error Handling](#error-handling-1)
- [Python Example](#python-example-13)
- [Limitations](#limitations)
- [Future Developments](#future-developments)
- [Conclusion](#conclusion-8)
- [Security Considerations](#security-considerations-2)
12. [Orchestration](#orchestration)
- [Key Concepts](#key-concepts)
- [Orchestration in R2R](#orchestration-in-r2r)
- [Benefits of Orchestration](#benefits-of-orchestration)
- [Workflows in R2R](#workflows-in-r2r)
- [List of Workflows](#list-of-workflows)
- [Orchestration GUI](#orchestration-gui)
- [Access GUI](#access-gui)
- [Login](#login-1)
- [Credentials](#credentials-1)
- [Logging into Hatchet](#logging-into-hatchet)
- [Running Tasks](#running-tasks)
- [Running Tasks Screenshot](#running-tasks-screenshot)
- [Inspecting a Workflow](#inspecting-a-workflow)
- [Inspecting a Workflow Screenshot](#inspecting-a-workflow-screenshot)
- [Long Running Tasks](#long-running-tasks)
- [Long Running Tasks Screenshot](#long-running-tasks-screenshot)
- [Coming Soon](#coming-soon)
- [Best Practices](#best-practices-6)
- [Development](#development-1)
- [Performance](#performance-2)
- [Quality](#quality-1)
- [Troubleshooting](#troubleshooting-3)
- [Conclusion](#conclusion-9)
13. [Maintenance & Scaling](#maintenance--scaling)
- [Vector Indices](#vector-indices)
- [Do You Need Vector Indices?](#do-you-need-vector-indices)
- [Vector Index Management](#vector-index-management)
- [Python Example: Creating and Deleting a Vector Index](#python-example-14)
- [Important Considerations](#important-considerations-1)
- [System Updates and Maintenance](#system-updates-and-maintenance)
- [Version Management](#version-management)
- [Check Current R2R Version](#check-current-r2r-version)
- [Update Process](#update-process)
- [Steps with Commands](#steps-with-commands)
- [Database Migration Management](#database-migration-management)
- [Check Current Migration](#check-current-migration)
- [Apply Migrations](#apply-migrations)
- [Managing Multiple Environments](#managing-multiple-environments)
- [Example with Environment Variables](#example-with-environment-variables)
- [Troubleshooting](#troubleshooting-4)
- [Steps](#steps-1)
- [Scaling Strategies](#scaling-strategies)
- [Horizontal Scaling](#horizontal-scaling)
- [Load Balancing](#load-balancing)
- [Sharding](#sharding)
- [Vertical Scaling](#vertical-scaling)
- [Cloud Provider Solutions](#cloud-provider-solutions)
- [Memory Optimization](#memory-optimization)
- [Multi-User Considerations](#multi-user-considerations)
- [Filtering Optimization](#filtering-optimization)
- [Collection Management](#collection-management-1)
- [Resource Allocation](#resource-allocation)
- [Performance Monitoring](#performance-monitoring)
- [Metrics](#metrics)
- [Performance Considerations](#performance-considerations-1)
- [Strategies](#strategies)
- [Additional Resources](#additional-resources-1)
- [Best Practices](#best-practices-7)
- [Optimize Indexing](#optimize-indexing)
- [Monitor Resources](#monitor-resources)
- [Regular Maintenance](#regular-maintenance)
- [Plan Scaling Ahead](#plan-scaling-ahead)
- [Conclusion](#conclusion-10)
14. [Web Development](#web-development)
- [Hello R2R—JavaScript](#hello-r2rjavascript)
- [Example: `r2r-js/examples/hello_r2r.js`](#example-r2r-jsexampleshello_r2rjs)
- [r2r-js Client](#r2r-js-client)
- [Installing](#installing-1)
- [Creating the Client](#creating-the-client)
- [Log into the Server](#log-into-the-server)
- [Ingesting Files](#ingesting-files-1)
- [Example and Sample Output](#example-and-sample-output-1)
- [Performing RAG](#performing-rag-1)
- [Example and Sample Output](#example-and-sample-output-2)
- [Connecting to a Web App](#connecting-to-a-web-app)
- [Setting up an API Route](#setting-up-an-api-route)
- [Frontend: React Component](#frontend-react-component)
- [Template Repository](#template-repository)
- [Usage Steps](#usage-steps-1)
- [Best Practices](#best-practices-8)
- [Secure API Routes](#secure-api-routes)
- [Optimize Frontend Performance](#optimize-frontend-performance)
- [Handle Errors Gracefully](#handle-errors-gracefully)
- [Implement Caching](#implement-caching)
- [Maintain Consistent State](#maintain-consistent-state)
- [Conclusion](#conclusion-11)
15. [User Management](#user-management)
- [Introduction](#introduction-3)
- [Basic Usage](#basic-usage-2)
- [User Registration and Login](#user-registration-and-login-1)
- [Python Example](#python-example-15)
- [Email Verification (Optional)](#email-verification-optional-1)
- [Token Refresh](#token-refresh-1)
- [User-Specific Search](#user-specific-search-1)
- [Curl Example](#curl-example-2)
- [User Logout](#user-logout-1)
- [Curl Example](#curl-example-3)
- [Advanced Authentication Features](#advanced-authentication-features-1)
- [Password Management](#password-management-1)
- [Python Example](#python-example-16)
- [User Profile Management](#user-profile-management-1)
- [Python Example](#python-example-17)
- [Account Deletion](#account-deletion-1)
- [Python Example](#python-example-18)
- [Logout](#logout-2)
- [Python Example](#python-example-19)
- [Superuser Capabilities and Default Admin Creation](#superuser-capabilities-and-default-admin-creation)
- [Superuser Capabilities](#superuser-capabilities-1)
- [Default Admin Creation](#default-admin-creation-1)
- [Configuration](#configuration-3)
- [Accessing Superuser Features](#accessing-superuser-features-1)
- [Python Example](#python-example-20)
- [Security Considerations for Superusers](#security-considerations-for-superusers)
- [Security Considerations](#security-considerations-3)
- [Customizing Authentication](#customizing-authentication)
- [Troubleshooting](#troubleshooting-5)
- [Conclusion](#conclusion-12)
16. [Collections](#collections)
- [Introduction](#introduction-4)
- [Basic Usage](#basic-usage-3)
- [Collection CRUD Operations](#collection-crud-operations-1)
- [Creating a Collection](#creating-a-collection)
- [Python Example](#python-example-21)
- [Retrieving Collection Details](#retrieving-collection-details)
- [Python Example](#python-example-22)
- [Updating a Collection](#updating-a-collection-1)
- [Python Example](#python-example-23)
- [Deleting a Collection](#deleting-a-collection-1)
- [Example](#example-13)
- [User Management in Collections](#user-management-in-collections)
- [Adding a User to a Collection](#adding-a-user-to-a-collection)
- [Example](#example-14)
- [Removing a User from a Collection](#removing-a-user-from-a-collection)
- [Example](#example-15)
- [Listing Users in a Collection](#listing-users-in-a-collection)
- [Example](#example-16)
- [Getting Collections for a User](#getting-collections-for-a-user)
- [Example](#example-17)
- [Document Management in Collections](#document-management-in-collections)
- [Assigning a Document to a Collection](#assigning-a-document-to-a-collection)
- [Example](#example-18)
- [Removing a Document from a Collection](#removing-a-document-from-a-collection)
- [Example](#example-19)
- [Listing Documents in a Collection](#listing-documents-in-a-collection)
- [Example](#example-20)
- [Getting Collections for a Document](#getting-collections-for-a-document)
- [Example](#example-21)
- [Advanced Collection Management](#advanced-collection-management)
- [Generating Synthetic Descriptions](#generating-synthetic-descriptions)
- [Example](#example-22)
- [Collection Overview](#collection-overview-1)
- [Example](#example-23)
- [Pagination and Filtering](#pagination-and-filtering-1)
- [Examples](#examples-1)
- [Security Considerations](#security-considerations-4)
- [Customizing Collection Permissions](#customizing-collection-permissions)
- [Troubleshooting](#troubleshooting-6)
- [Conclusion](#conclusion-13)
- [Next Steps](#next-steps-3)
17. [Telemetry](#telemetry)
- [Introduction](#introduction-5)
- [Disabling Telemetry](#disabling-telemetry)
- [Example](#example-24)
- [Collected Information](#collected-information)
- [Telemetry Data Storage](#telemetry-data-storage)
- [Note](#note)
- [Why We Collect Telemetry](#why-we-collect-telemetry)
- [Conclusion](#conclusion-14)
18. [Embedding](#embedding)
- [Embedding System](#embedding-system)
- [Embedding Configuration](#embedding-configuration-1)
- [Example: `r2r.toml`](#example-r2rtoml-1)
- [Advanced Embedding Features in R2R](#advanced-embedding-features-in-r2r)
- [Batched Processing](#batched-processing)
- [Python Example](#python-example-24)
- [Concurrent Request Management](#concurrent-request-management-1)
- [Performance Considerations](#performance-considerations-2)
- [Strategies](#strategies-1)
- [Supported LiteLLM Providers](#supported-litellm-providers)
- [Example Configuration](#example-configuration-9)
- [Supported Models](#supported-models)
- [Performance Considerations](#performance-considerations-3)
- [Conclusion](#conclusion-15)
19. [Prompts](#prompts)
- [Prompt Management in R2R](#prompt-management-in-r2r)
- [Default Prompts](#default-prompts)
- [Example: `rag.yaml`](#example-default_ragyaml)
- [Prompt Files](#prompt-files)
- [Prompt Provider](#prompt-provider)
- [Prompt Structure](#prompt-structure)
- [Managing Prompts](#managing-prompts)
- [Adding a Prompt](#adding-a-prompt)
- [Example](#example-25)
- [Updating a Prompt](#updating-a-prompt)
- [Example](#example-26)
- [Retrieving a Prompt](#retrieving-a-prompt)
- [Example](#example-27)
- [Security Considerations](#security-considerations-5)
- [Conclusion](#conclusion-16)
20. [RAG](#rag)
- [RAG Customization](#rag-customization)
- [Components](#components)
- [LLM Provider Configuration](#llm-provider-configuration)
- [Retrieval Configuration](#retrieval-configuration-1)
- [Combining LLM and Retrieval Configuration for RAG](#combining-llm-and-retrieval-configuration-for-rag)
- [Example](#example-28)
- [RAG Prompt Override](#rag-prompt-override)
- [Example](#example-29)
- [Agent-based Interaction](#agent-based-interaction)
- [Example](#example-30)
- [Conclusion](#conclusion-17)
21. [Graphs](#graphs)
- [Graphs](#graphs-1)
- [Knowledge Graph Operations](#knowledge-graph-operations)
- [Entity Management](#entity-management-1)
- [Relationship Management](#relationship-management-1)
- [Batch Import](#batch-import)
- [Vector Search](#vector-search-1)
- [Community Detection](#community-detection)
- [Customization](#customization-1)
- [Conclusion](#conclusion-18)
22. [Conclusion](#conclusion-19)
---
## Introduction
**R2R** (Retrieval to Riches) is an engine for building user-facing **Retrieval-Augmented Generation (RAG)** applications. It provides core services through an architecture of providers, services, and an integrated RESTful API. This documentation offers a detailed walkthrough of interacting with R2R, including installation, configuration, and leveraging its advanced features such as data ingestion, search, RAG, and knowledge graphs.
For a deeper dive into the R2R system architecture, refer to the [R2R System Architecture](https://r2r-docs.sciphi.ai/introduction/system).
---
## Installation
Before diving into R2R's features, ensure that you have completed the [installation instructions](https://r2r-docs.sciphi.ai/documentation/installation/overview).
### Prerequisites
- **Python 3.8+**: Ensure Python is installed on your system.
- **Docker**: Required for Docker-based installations. Install Docker from the [official Docker installation guide](https://docs.docker.com/engine/install/).
- **pip**: Python package installer.
### Docker Installation
This installation guide is for the **Full R2R**. For solo developers or teams prototyping, start with [R2R Light](https://r2r-docs.sciphi.ai/documentation/installation/light/local-system).
#### Install the R2R CLI & Python SDK
```bash
pip install r2r
```
> **Note**: A distinct CLI binary for R2R is under active development. For specific needs or feature requests, reach out to the R2R team.
#### Start R2R with Docker
The Full R2R installation uses a custom configuration (`full.toml`). Launch R2R with Docker:
```bash
r2r serve --docker --config-path=full.toml
```
> This command pulls necessary Docker images and starts required containers, including R2R, Hatchet, and Postgres+pgvector. Access the live server at [http://localhost:7272](http://localhost:7272/).
### Google Cloud Platform Deployment
Deploying R2R on Google Cloud Platform (GCP) involves setting up a Compute Engine instance, installing dependencies, and configuring port forwarding.
#### Overview
1. **Creating a Google Compute Engine Instance**
2. **Installing Dependencies**
3. **Setting up R2R**
4. **Configuring Port Forwarding for Local Access**
5. **Exposing Ports for Public Access (Optional)**
6. **Security Considerations**
#### Creating a Google Compute Engine Instance
1. **Log in** to the Google Cloud Console.
2. Navigate to **Compute Engine** > **VM instances**.
3. Click **Create Instance**.
4. Configure the instance:
- **Name**: Choose a name.
- **Region and Zone**: Select based on preference.
- **Machine Configuration**:
- **Series**: N1
- **Machine type**: `n1-standard-4` (4 vCPU, 15 GB memory) or higher.
- **Boot Disk**:
- **OS**: Ubuntu 22.04 LTS
- **Size**: 500 GB
- **Firewall**: Allow HTTP and HTTPS traffic.
5. Click **Create** to launch the instance.
#### Installing Dependencies
SSH into your instance and run the following commands:
```bash
# Update package list and install Python and pip
sudo apt update
sudo apt install python3-pip -y
# Install R2R
pip install r2r
# Add R2R to PATH
echo 'export PATH=$PATH:$HOME/.local/bin' >> ~/.bashrc
source ~/.bashrc
# Install Docker
sudo apt-get update
sudo apt-get install ca-certificates curl gnupg -y
sudo install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
sudo chmod a+r /etc/apt/keyrings/docker.gpg
echo \
"deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
$(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin -y
# Add your user to the Docker group
sudo usermod -aG docker $USER
newgrp docker
# Verify Docker installation
docker run hello-world
```
#### Setting up R2R
```bash
# Set required remote providers
export OPENAI_API_KEY=sk-...
# Optional - pass in a custom configuration
r2r serve --docker --full
```
#### Configuring Port Forwarding for Local Access
Use SSH port forwarding to access R2R locally:
```bash
gcloud compute ssh --zone "your-zone" "your-instance-name" -- -L 7273:localhost:7273 -L 7274:localhost:7274
```
#### Exposing Ports for Public Access (Optional)
To make R2R publicly accessible:
1. **Create a Firewall Rule**:
- Navigate to **VPC network** > **Firewall**.
- Click **Create Firewall Rule**.
- **Name**: Allow-R2R
- **Target tags**: `r2r-server`
- **Source IP ranges**: `0.0.0.0/0`
- **Protocols and ports**: `tcp:7272`
2. **Add Network Tag to Instance**:
- Go to **Compute Engine** > **VM instances**.
- Click on your instance.
- Click **Edit**.
- Under **Network tags**, add `r2r-server`.
- Click **Save**.
3. **Ensure R2R Listens on All Interfaces**.
After starting R2R, access it at:
```
http://<your-instance-external-ip>:7272
```
> **Security Considerations**:
> - Use HTTPS with a valid SSL certificate.
> - Restrict source IP addresses in firewall rules.
> - Regularly update and patch your system.
#### Conclusion
You have successfully deployed R2R on Google Cloud Platform. The application is accessible locally via SSH tunneling and optionally publicly. Ensure proper security measures are in place before exposing R2R to the internet.
For more details, refer to the [R2R Configuration Documentation](https://r2r-docs.sciphi.ai/documentation/configuration/overview).
---
## R2R Application Lifecycle
R2R's application lifecycle encompasses customization, configuration, deployment, implementation, and interaction. The lifecycle is designed to provide flexibility and scalability for various use cases.
### Developer Workflow
- **Customize**: Developers tailor R2R applications using R2RConfig and the R2R SDK.
- **Configure**: Adjust settings via configuration files (`r2r.toml`) or runtime overrides.
- **Deploy**: Launch R2R using Docker, cloud platforms, or local installations.
- **Implement**: Integrate R2R into applications using provided APIs and SDKs.
- **Interact**: Users engage with the R2R application through interfaces like dashboards or APIs to perform RAG queries or search documents.
### User Interaction
- **Users** interact with the R2R application, typically over an HTTP interface, to run RAG queries or search documents.
- Access the **R2R Dashboard** for managing documents, collections, and performing searches.
### Hello R2R (Code Example)
**Python Example** at `core/examples/hello_r2r.py`:
```python
from r2r import R2RClient
client = R2RClient("http://localhost:7272")
# Create a test document
with open("test.txt", "w") as file:
file.write("John is a person that works at Google.")
client.documents.create(file_path="test.txt")
# Call RAG directly
rag_response = client.retrieval.rag(
query="Who is John",
rag_generation_config={"model": "openai/gpt-4.1-mini", "temperature": 0.0},
)
results = rag_response["results"]
print(f"Search Results:\n{results['search_results']}")
print(f"Completion:\n{results['completion']}")
```
**Sample Output:**
```json
{
"results": {
"search_results": {
"chunk_search_results": [
{
"chunk_id": "b9f40dbd-2c8e-5c0a-8454-027ac45cb0ed",
"document_id": "7c319fbe-ca61-5770-bae2-c3d0eaa8f45c",
"score": 0.6847735847465275,
"text": "John is a person that works at Google.",
"metadata": {
"version": "v0",
"chunk_order": 0,
"document_type": "txt",
"associated_query": "Who is John"
}
}
],
"kg_search_results": []
},
"completion": {
"id": "chatcmpl-AV1Sc9DORfHvq7yrmukxfJPDV5dCB",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"message": {
"content": "John is a person that works at Google [1].",
"role": "assistant"
}
}
],
"created": 1731957146,
"model": "gpt-4.1-mini",
"object": "chat.completion",
"usage": {
"completion_tokens": 11,
"prompt_tokens": 145,
"total_tokens": 156
}
}
}
}
```
This snippet:
1. Creates a file with simple text.
2. Ingests it to R2R.
3. Runs a **Retrieval-Augmented Generation** query.
4. Prints the context matched (“search_results”) and the generated answer (“completion”).
---
## Configuration
R2R is highly configurable, allowing you to tailor its behavior to your specific needs. Configuration can be done at the server-side using configuration files (`r2r.toml`) or at runtime via API calls.
### Configuration Overview
R2R configurations are divided into two primary levels:
1. **Server-Side Configuration**: Managed through the `r2r.toml` file and environment variables.
2. **Runtime Overrides**: Passed directly in API calls to adjust settings dynamically.
### Server-Side Configuration (`r2r.toml`)
The `r2r.toml` file allows you to define server-side settings that govern the behavior of R2R. Below are the main configuration sections:
#### Example: `r2r.toml`
```toml
[completion]
provider = "litellm"
concurrent_request_limit = 16
[completion.generation_config]
model = "openai/gpt-4.1"
temperature = 0.5
[ingestion]
provider = "r2r"
chunking_strategy = "recursive"
chunk_size = 1024
chunk_overlap = 512
excluded_parsers = ["mp4"]
[database]
provider = "postgres"
user = "your_postgres_user"
password = "your_postgres_password"
host = "your_postgres_host"
port = "your_postgres_port"
db_name = "your_database_name"
project_name = "your_project_name"
[embedding]
provider = "litellm"
base_model = "openai/text-embedding-3-small"
base_dimension = 512
batch_size = 512
rerank_model = "BAAI/bge-reranker-v2-m3"
concurrent_request_limit = 256
[auth]
provider = "r2r"
require_authentication = true
require_email_verification = false
default_admin_email = "admin@example.com"
default_admin_password = "change_me_immediately"
access_token_lifetime_in_minutes = 60
refresh_token_lifetime_in_days = 7
secret_key = "your-secret-key"
[ingestion.chunk_enrichment_settings]
enable_chunk_enrichment = true
strategies = ["semantic", "neighborhood"]
forward_chunks = 3
backward_chunks = 3
semantic_neighbors = 10
semantic_similarity_threshold = 0.7
generation_config = { model = "openai/gpt-4.1-mini" }
[agent]
rag_agent_static_prompt = "rag_agent"
tools = ["search_file_knowledge", "web_search"]
[database.graph_creation_settings]
entity_types = []
relation_types = []
max_knowledge_triples = 100
fragment_merge_count = 4
generation_config = { model = "openai/gpt-4.1-mini" }
[database.graph_enrichment_settings]
max_description_input_length = 65536
max_summary_input_length = 65536
generation_config = { model = "openai/gpt-4.1-mini" }
leiden_params = {}
[database.graph_settings]
generation_config = { model = "openai/gpt-4.1-mini" }
```
### Runtime Overrides
Runtime overrides allow you to adjust configurations dynamically without modifying the `r2r.toml` file. This is useful for temporary changes or testing different settings on the fly.
**Example: Customizing RAG Query at Runtime**
```python
rag_response = client.retrieval.rag(
query="Who is Jon Snow?",
rag_generation_config={
"model": "anthropic/claude-3-haiku-20240307",
"temperature": 0.7
},
search_settings={
"use_semantic_search": True,
"limit": 20,
"use_hybrid_search": True
}
)
```
### Postgres Configuration
R2R uses Postgres for relational and vector data storage, leveraging the `pgvector` extension for vector indexing.
#### Example Configuration
```toml
[database]
provider = "postgres"
user = "your_postgres_user"
password = "your_postgres_password"
host = "your_postgres_host"
port = "your_postgres_port"
db_name = "your_database_name"
project_name = "your_project_name"
```
**Key Features:**
- **pgvector**: Enables efficient vector operations.
- **Full-Text Indexing**: Utilizes Postgres’s `ts_rank` for full-text search.
- **JSONB**: Stores flexible metadata.
### Embedding Configuration
R2R uses **LiteLLM** to manage embedding providers, allowing flexibility in selecting different LLM providers.
#### Example Configuration
```toml
[embedding]
provider = "litellm"
base_model = "openai/text-embedding-3-small"
base_dimension = 512
batch_size = 512
rerank_model = "BAAI/bge-reranker-v2-m3"
concurrent_request_limit = 256
```
**Environment Variables:**
- `OPENAI_API_KEY`
- `HUGGINGFACE_API_KEY`
- `ANTHROPIC_API_KEY`
- `COHERE_API_KEY`
- `OLLAMA_API_KEY`
- etc.
**Supported Providers:**
- OpenAI
- Azure
- Anthropic
- Cohere
- Ollama
- HuggingFace
- Bedrock
- Vertex AI
- Voyage AI
### Auth & Users Configuration
R2R’s authentication system supports secure user registration, login, session management, and access control.
#### Example Configuration
```toml
[auth]
provider = "r2r"
require_authentication = true
require_email_verification = false
default_admin_email = "admin@example.com"
default_admin_password = "change_me_immediately"
access_token_lifetime_in_minutes = 60
refresh_token_lifetime_in_days = 7
secret_key = "your-secret-key"
```
**Key Features:**
- **JWT-Based Authentication**: Utilizes access and refresh tokens.
- **Email Verification**: Optional, recommended for production.
- **Superuser Management**: Default admin creation and superuser capabilities.
### Data Ingestion Configuration
Configure how R2R ingests documents, including parsing, chunking, and embedding strategies.
#### Example Configuration
```toml
[ingestion]
provider = "r2r"
chunking_strategy = "recursive"
chunk_size = 1024
chunk_overlap = 512
excluded_parsers = ["mp4"]
[ingestion.chunk_enrichment_settings]
enable_chunk_enrichment = true
strategies = ["semantic", "neighborhood"]
forward_chunks = 3
backward_chunks = 3
semantic_neighbors = 10
semantic_similarity_threshold = 0.7
generation_config = { model = "openai/gpt-4.1-mini" }
```
**Modes:**
- `fast`: Speed-oriented ingestion.
- `hi-res`: Comprehensive, high-quality ingestion.
- `custom`: Fine-grained control with a full `ingestion_config` dictionary.
### Retrieval Configuration
Focuses on search settings, combining vector and knowledge-graph search capabilities.
#### Example Configuration
```json
{
"search_settings": {
"use_semantic_search": true,
"limit": 20,
"use_hybrid_search": true,
"graph_search_settings": {
"use_graph_search": true,
"kg_search_type": "local"
}
}
}
```
### RAG Configuration
Customize RAG (Retrieval-Augmented Generation) settings, including the language model's behavior.
#### Example Configuration
```python
rag_generation_config = {
"model": "anthropic/claude-3-haiku-20240307",
"temperature": 0.7,
"top_p": 0.95,
"max_tokens_to_sample": 1500,
"stream": True
}
```
### Graphs Configuration
Defines settings related to knowledge graph creation and enrichment.
#### Example Configuration
```toml
[database.graph_creation_settings]
entity_types = []
relation_types = []
max_knowledge_triples = 100
fragment_merge_count = 4
generation_config = { model = "openai/gpt-4.1-mini" }
[database.graph_enrichment_settings]
max_description_input_length = 65536
max_summary_input_length = 65536
generation_config = { model = "openai/gpt-4.1-mini" }
leiden_params = {}
[database.graph_settings]
generation_config = { model = "openai/gpt-4.1-mini" }
```
### Prompts Configuration
Manages prompt templates used for various tasks within R2R.
#### Example Configuration
Prompts are stored in Postgres and can be managed via the SDK.
**Example: Adding a Prompt**
```python
response = client.prompts.add_prompt(
name="my_new_prompt",
template="Hello, {name}! Welcome to {service}.",
input_types={"name": "str", "service": "str"}
)
```
---
## Data Ingestion
### Introduction
R2R provides a powerful and flexible ingestion pipeline to process and manage various types of documents. It supports a wide range of file formats—text, documents, PDFs, images, audio, and video—and transforms them into searchable, analyzable content. The ingestion process includes parsing, chunking, embedding, and optionally extracting entities and relationships for knowledge graph construction.