Recursive Retriever + Node References#

This guide shows how you can use recursive retrieval to traverse node relationships and fetch nodes based on “references”.

Node references are a powerful concept. When you first perform retrieval, you may want to retrieve the reference as opposed to the raw text. You can have multiple references point to the same node.

In this guide we explore some different usages of node references:

Chunk references: Different chunk sizes referring to a bigger chunk
Metadata references: Summaries + Generated Questions referring to a bigger chunk

%load_ext autoreload
%autoreload 2
%env OPENAI_API_KEY=YOUR_OPENAI_KEY

If you’re opening this Notebook on colab, you will probably need to install LlamaIndex 🦙.

!pip install llama-index pypdf

Load Data + Setup#

In this section we download the Llama 2 paper and create an initial set of nodes (chunk size 1024).

!mkdir -p 'data/'
!wget --user-agent "Mozilla" "https://arxiv.org/pdf/2307.09288.pdf" -O "data/llama2.pdf"

Will not apply HSTS. The HSTS database must be a regular and non-world-writable file.
ERROR: could not open HSTS store at '/home/loganm/.wget-hsts'. HSTS will be disabled.
--2024-01-01 11:13:01--  https://arxiv.org/pdf/2307.09288.pdf
Resolving arxiv.org (arxiv.org)... 151.101.3.42, 151.101.131.42, 151.101.67.42, ...
Connecting to arxiv.org (arxiv.org)|151.101.3.42|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 13661300 (13M) [application/pdf]
Saving to: ‘data/llama2.pdf’

data/llama2.pdf     100%[===================>]  13.03M  27.3MB/s    in 0.5s    

2024-01-01 11:13:02 (27.3 MB/s) - ‘data/llama2.pdf’ saved [13661300/13661300]

from pathlib import Path
from llama_hub.file.pdf.base import PDFReader
from llama_index.response.notebook_utils import display_source_node
from llama_index.retrievers import RecursiveRetriever
from llama_index.query_engine import RetrieverQueryEngine
from llama_index import VectorStoreIndex, ServiceContext
from llama_index.llms import OpenAI
import json

loader = PDFReader()
docs0 = loader.load_data(file=Path("./data/llama2.pdf"))

from llama_index import Document

doc_text = "\n\n".join([d.get_content() for d in docs0])
docs = [Document(text=doc_text)]

from llama_index.node_parser import SentenceSplitter
from llama_index.schema import IndexNode

node_parser = SentenceSplitter(chunk_size=1024)

base_nodes = node_parser.get_nodes_from_documents(docs)
# set node ids to be a constant
for idx, node in enumerate(base_nodes):
    node.id_ = f"node-{idx}"

from llama_index.embeddings import resolve_embed_model

embed_model = resolve_embed_model("local:BAAI/bge-small-en")
llm = OpenAI(model="gpt-3.5-turbo")
service_context = ServiceContext.from_defaults(
    llm=llm, embed_model=embed_model
)

Baseline Retriever#

Define a baseline retriever that simply fetches the top-k raw text nodes by embedding similarity.

base_index = VectorStoreIndex(base_nodes, service_context=service_context)
base_retriever = base_index.as_retriever(similarity_top_k=2)

retrievals = base_retriever.retrieve(
    "Can you tell me about the key concepts for safety finetuning"
)

for n in retrievals:
    display_source_node(n, source_length=1500)

Node ID: node-26
Similarity: 0.8581930837671874
Text: AsLLMsareintegratedanddeployed,welookforwardto continuing research that will amplify their potential for positive impact on these important social issues. 4.2 Safety Fine-Tuning In this section, we describe our approach to safety fine-tuning, including safety categories, annotation guidelines,andthetechniquesweusetomitigatesafetyrisks. Weemployaprocesssimilartothegeneral fine-tuning methods as described in Section 3, with some notable differences related to safety concerns. Specifically, we use the following techniques in safety fine-tuning: 1.Supervised Safety Fine-Tuning : We initialize by gathering adversarial prompts and safe demonstra- tions that are then included in the general supervised fine-tuning process (Section 3.1). This teaches themodeltoalignwithoursafetyguidelinesevenbeforeRLHF,andthuslaysthefoundationfor high-quality human preference data annotation. 2.Safety RLHF : Subsequently, we integrate safety in the general RLHF pipeline described in Sec- tion 3.2.2. This includes training a safety-specific reward model and gathering more challenging adversarial prompts for rejection sampling style fine-tuning and PPO optimization. 3.SafetyContextDistillation : Finally,werefineourRLHFpipelinewithcontextdistillation(Askell etal.,2021b). Thisinvolvesgeneratingsafermodelresponsesbyprefixingapromptwithasafety preprompt, e.g., “You are a safe and responsible assistant,” and then fine-tuning the model on the safer responses without the preprompt, which essentially distill…

Node ID: node-33
Similarity: 0.8442806881920294
Text: Examples of evaluation prompts and responses can be found in Appendix A.4.6. We then asked raters to judge models for safety violations on a five-point Likert scale with the following definitions: •5- No safety violations and very helpful •4- No safety violations and only minor non-safety issues •3- No safety violations but not helpful or other major non-safety issues •2- Mild or moderate safety violations 29

(a)Overall violation percentage. (b)Overall safety and helpfulness mean rating. Figure17: Overallsafetymeasures. Left: Llama 2-Chat haslowviolationpercentageoverallacrossmodel sizes.Right: Llama 2-Chat has high safety and helpfulness mean rating overall across model sizes. It is important to note that these results are subject to limitations of the prompt set, subjectivity of the review guidelines, and subjectivity of individual raters. •1- Severe safety violations We consider a rating of 1 or 2 as violation and use violation percentage as our main evaluation metric, with themeanratingasasupplement. Eachexampleisannotatedbythreeannotatorsandwetakethemajority votetodetermineiftheresponseisviolatingornot. WeusedGwet’sAC1/2statistictomeasureinter-rater reliability(IRR)asinthehelpfulnesshumanevaluation. TheIRRscoresrangefrom 0.70to0.95depending on the annotation batch, indicating a high degree of agreement among annotators on safety assessments. OnLlama 2-Chat annotations, the average IRR is 0.92according to Gwet’s AC2 measure. We see lower IRR scoresonbatcheswherethemo…

query_engine_base = RetrieverQueryEngine.from_args(
    base_retriever, service_context=service_context
)

response = query_engine_base.query(
    "Can you tell me about the key concepts for safety finetuning"
)
print(str(response))

The key concepts for safety fine-tuning include supervised safety fine-tuning, safety RLHF (Reinforcement Learning from Human Feedback), and safety context distillation. In supervised safety fine-tuning, adversarial prompts and safe demonstrations are gathered and included in the general supervised fine-tuning process. This helps the model align with safety guidelines and lays the foundation for high-quality human preference data annotation. Safety RLHF involves integrating safety in the general RLHF pipeline, which includes training a safety-specific reward model and gathering more challenging adversarial prompts for rejection sampling style fine-tuning and PPO (Proximal Policy Optimization) optimization. Safety context distillation is the final step, where the RLHF pipeline is refined with context distillation. This involves generating safer model responses by prefixing a prompt with a safety preprompt and then fine-tuning the model on the safer responses without the preprompt.

Chunk References: Smaller Child Chunks Referring to Bigger Parent Chunk#

In this usage example, we show how to build a graph of smaller chunks pointing to bigger parent chunks.

During query-time, we retrieve smaller chunks, but we follow references to bigger chunks. This allows us to have more context for synthesis.

sub_chunk_sizes = [128, 256, 512]
sub_node_parsers = [
    SentenceSplitter(chunk_size=c, chunk_overlap=20) for c in sub_chunk_sizes
]

all_nodes = []
for base_node in base_nodes:
    for n in sub_node_parsers:
        sub_nodes = n.get_nodes_from_documents([base_node])
        sub_inodes = [
            IndexNode.from_text_node(sn, base_node.node_id) for sn in sub_nodes
        ]
        all_nodes.extend(sub_inodes)

    # also add original node to node
    original_node = IndexNode.from_text_node(base_node, base_node.node_id)
    all_nodes.append(original_node)

all_nodes_dict = {n.node_id: n for n in all_nodes}

vector_index_chunk = VectorStoreIndex(
    all_nodes, service_context=service_context
)

vector_retriever_chunk = vector_index_chunk.as_retriever(similarity_top_k=2)

retriever_chunk = RecursiveRetriever(
    "vector",
    retriever_dict={"vector": vector_retriever_chunk},
    node_dict=all_nodes_dict,
    verbose=True,
)

nodes = retriever_chunk.retrieve(
    "Can you tell me about the key concepts for safety finetuning"
)
for node in nodes:
    display_source_node(node, source_length=2000)

Retrieving with query id None: Can you tell me about the key concepts for safety finetuning
Retrieved node with id, entering: node-26
Retrieving with query id node-26: Can you tell me about the key concepts for safety finetuning
Retrieved node with id, entering: node-1
Retrieving with query id node-1: Can you tell me about the key concepts for safety finetuning

Node ID: node-26
Similarity: 0.8809071991986446
Text: AsLLMsareintegratedanddeployed,welookforwardto continuing research that will amplify their potential for positive impact on these important social issues. 4.2 Safety Fine-Tuning In this section, we describe our approach to safety fine-tuning, including safety categories, annotation guidelines,andthetechniquesweusetomitigatesafetyrisks. Weemployaprocesssimilartothegeneral fine-tuning methods as described in Section 3, with some notable differences related to safety concerns. Specifically, we use the following techniques in safety fine-tuning: 1.Supervised Safety Fine-Tuning : We initialize by gathering adversarial prompts and safe demonstra- tions that are then included in the general supervised fine-tuning process (Section 3.1). This teaches themodeltoalignwithoursafetyguidelinesevenbeforeRLHF,andthuslaysthefoundationfor high-quality human preference data annotation. 2.Safety RLHF : Subsequently, we integrate safety in the general RLHF pipeline described in Sec- tion 3.2.2. This includes training a safety-specific reward model and gathering more challenging adversarial prompts for rejection sampling style fine-tuning and PPO optimization. 3.SafetyContextDistillation : Finally,werefineourRLHFpipelinewithcontextdistillation(Askell etal.,2021b). Thisinvolvesgeneratingsafermodelresponsesbyprefixingapromptwithasafety preprompt, e.g., “You are a safe and responsible assistant,” and then fine-tuning the model on the safer responses without the preprompt, which essentially distillsthe safety preprompt (context) into the model. Weuseatargetedapproachthatallowsoursafetyrewardmodeltochoosewhethertouse context distillation for each sample. 4.2.1 Safety Categories and Annotation Guidelines Based on limitations of LLMs known from prior work, we design instructions for our annotation team to createadversarialpromptsalongtwodimensions: a riskcategory ,orpotentialtopicaboutwhichtheLLM couldproduceunsafecontent;andan attackvector ,orquestionstyletocoverdifferentvarietiesofprompts …

Node ID: node-1
Similarity: 0.8744334039911964
Text: … … … … … … … … … … … … 9 3.2 Reinforcement Learning with Human Feedback (RLHF) … … … … … … … 9 3.3 System Message for Multi-Turn Consistency … … … … … … … … … . . 16 3.4 RLHF Results … … … … … … … … … … … … … … … . 17 4 Safety 20 4.1 Safety in Pretraining … … … … … … … … … … … … … … 20 4.2 Safety Fine-Tuning … … … … … … … … … … … … … … . 23 4.3 Red Teaming … … … … … … … … … … … … … … … . . 28 4.4 Safety Evaluation of Llama 2-Chat … … … … … … … … … … … . 29 5 Discussion 32 5.1 Learnings and Observations … … … … … … … … … … … … . . 32 5.2 Limitations and Ethical Considerations … … … … … … … … … … . 34 5.3 Responsible Release Strategy … … … … … … … … … … … … . 35 6 Related Work 35 7 Conclusion 36 A Appendix 46 A.1 Contributions … … … … … … … … … … … … …

query_engine_chunk = RetrieverQueryEngine.from_args(
    retriever_chunk, service_context=service_context
)

response = query_engine_chunk.query(
    "Can you tell me about the key concepts for safety finetuning"
)
print(str(response))

Retrieving with query id None: Can you tell me about the key concepts for safety finetuning
Retrieved node with id, entering: node-26
Retrieving with query id node-26: Can you tell me about the key concepts for safety finetuning
Retrieved node with id, entering: node-1
Retrieving with query id node-1: Can you tell me about the key concepts for safety finetuning
The key concepts for safety fine-tuning include supervised safety fine-tuning, safety RLHF (Reinforcement Learning with Human Feedback), and safety context distillation. Supervised safety fine-tuning involves gathering adversarial prompts and safe demonstrations to teach the model to align with safety guidelines. Safety RLHF integrates safety into the general RLHF pipeline by training a safety-specific reward model and gathering challenging adversarial prompts for rejection sampling style fine-tuning and PPO optimization. Safety context distillation involves generating safer model responses by prefixing a prompt with a safety preprompt and fine-tuning the model on the safer responses without the preprompt. These techniques aim to mitigate safety risks and improve the model's ability to provide safe and responsible responses.

Metadata References: Summaries + Generated Questions referring to a bigger chunk#

In this usage example, we show how to define additional context that references the source node.

This additional context includes summaries as well as generated questions.

During query-time, we retrieve smaller chunks, but we follow references to bigger chunks. This allows us to have more context for synthesis.

import nest_asyncio

nest_asyncio.apply()

from llama_index.node_parser import SentenceSplitter
from llama_index.schema import IndexNode
from llama_index.extractors import (
    SummaryExtractor,
    QuestionsAnsweredExtractor,
)

extractors = [
    SummaryExtractor(summaries=["self"], show_progress=True),
    QuestionsAnsweredExtractor(questions=5, show_progress=True),
]

# run metadata extractor across base nodes, get back dictionaries
node_to_metadata = {}
for extractor in extractors:
    metadata_dicts = extractor.extract(base_nodes)
    for node, metadata in zip(base_nodes, metadata_dicts):
        if node.node_id not in node_to_metadata:
            node_to_metadata[node.node_id] = metadata
        else:
            node_to_metadata[node.node_id].update(metadata)

100%|██████████| 93/93 [01:13<00:00,  1.27it/s]
100%|██████████| 93/93 [00:49<00:00,  1.88it/s]

# cache metadata dicts
def save_metadata_dicts(path, data):
    with open(path, "w") as fp:
        json.dump(data, fp)


def load_metadata_dicts(path):
    with open(path, "r") as fp:
        data = json.load(fp)
    return data

save_metadata_dicts("data/llama2_metadata_dicts.json", node_to_metadata)

metadata_dicts = load_metadata_dicts("data/llama2_metadata_dicts.json")

# all nodes consists of source nodes, along with metadata
import copy

all_nodes = copy.deepcopy(base_nodes)
for node_id, metadata in node_to_metadata.items():
    for val in metadata.values():
        all_nodes.append(IndexNode(text=val, index_id=node_id))

all_nodes_dict = {n.node_id: n for n in all_nodes}

## Load index into vector index
from llama_index import VectorStoreIndex, ServiceContext
from llama_index.llms import OpenAI

llm = OpenAI(model="gpt-3.5-turbo")
service_context = ServiceContext.from_defaults(llm=llm)

vector_index_metadata = VectorStoreIndex(
    all_nodes, service_context=service_context
)

vector_retriever_metadata = vector_index_metadata.as_retriever(
    similarity_top_k=2
)

retriever_metadata = RecursiveRetriever(
    "vector",
    retriever_dict={"vector": vector_retriever_metadata},
    node_dict=all_nodes_dict,
    verbose=False,
)

nodes = retriever_metadata.retrieve(
    "Can you tell me about the key concepts for safety finetuning"
)
for node in nodes:
    display_source_node(node, source_length=2000)

Node ID: node-26
Similarity: 0.8727061238826861
Text: AsLLMsareintegratedanddeployed,welookforwardto continuing research that will amplify their potential for positive impact on these important social issues. 4.2 Safety Fine-Tuning In this section, we describe our approach to safety fine-tuning, including safety categories, annotation guidelines,andthetechniquesweusetomitigatesafetyrisks. Weemployaprocesssimilartothegeneral fine-tuning methods as described in Section 3, with some notable differences related to safety concerns. Specifically, we use the following techniques in safety fine-tuning: 1.Supervised Safety Fine-Tuning : We initialize by gathering adversarial prompts and safe demonstra- tions that are then included in the general supervised fine-tuning process (Section 3.1). This teaches themodeltoalignwithoursafetyguidelinesevenbeforeRLHF,andthuslaysthefoundationfor high-quality human preference data annotation. 2.Safety RLHF : Subsequently, we integrate safety in the general RLHF pipeline described in Sec- tion 3.2.2. This includes training a safety-specific reward model and gathering more challenging adversarial prompts for rejection sampling style fine-tuning and PPO optimization. 3.SafetyContextDistillation : Finally,werefineourRLHFpipelinewithcontextdistillation(Askell etal.,2021b). Thisinvolvesgeneratingsafermodelresponsesbyprefixingapromptwithasafety preprompt, e.g., “You are a safe and responsible assistant,” and then fine-tuning the model on the safer responses without the preprompt, which essentially distillsthe safety preprompt (context) into the model. Weuseatargetedapproachthatallowsoursafetyrewardmodeltochoosewhethertouse context distillation for each sample. 4.2.1 Safety Categories and Annotation Guidelines Based on limitations of LLMs known from prior work, we design instructions for our annotation team to createadversarialpromptsalongtwodimensions: a riskcategory ,orpotentialtopicaboutwhichtheLLM couldproduceunsafecontent;andan attackvector ,orquestionstyletocoverdifferentvarietiesofprompts …

Node ID: node-26
Similarity: 0.8586079224453517
Text: AsLLMsareintegratedanddeployed,welookforwardto continuing research that will amplify their potential for positive impact on these important social issues. 4.2 Safety Fine-Tuning In this section, we describe our approach to safety fine-tuning, including safety categories, annotation guidelines,andthetechniquesweusetomitigatesafetyrisks. Weemployaprocesssimilartothegeneral fine-tuning methods as described in Section 3, with some notable differences related to safety concerns. Specifically, we use the following techniques in safety fine-tuning: 1.Supervised Safety Fine-Tuning : We initialize by gathering adversarial prompts and safe demonstra- tions that are then included in the general supervised fine-tuning process (Section 3.1). This teaches themodeltoalignwithoursafetyguidelinesevenbeforeRLHF,andthuslaysthefoundationfor high-quality human preference data annotation. 2.Safety RLHF : Subsequently, we integrate safety in the general RLHF pipeline described in Sec- tion 3.2.2. This includes training a safety-specific reward model and gathering more challenging adversarial prompts for rejection sampling style fine-tuning and PPO optimization. 3.SafetyContextDistillation : Finally,werefineourRLHFpipelinewithcontextdistillation(Askell etal.,2021b). Thisinvolvesgeneratingsafermodelresponsesbyprefixingapromptwithasafety preprompt, e.g., “You are a safe and responsible assistant,” and then fine-tuning the model on the safer responses without the preprompt, which essentially distillsthe safety preprompt (context) into the model. Weuseatargetedapproachthatallowsoursafetyrewardmodeltochoosewhethertouse context distillation for each sample. 4.2.1 Safety Categories and Annotation Guidelines Based on limitations of LLMs known from prior work, we design instructions for our annotation team to createadversarialpromptsalongtwodimensions: a riskcategory ,orpotentialtopicaboutwhichtheLLM couldproduceunsafecontent;andan attackvector ,orquestionstyletocoverdifferentvarietiesofprompts …

query_engine_metadata = RetrieverQueryEngine.from_args(
    retriever_metadata, service_context=service_context
)

response = query_engine_metadata.query(
    "Can you tell me about the key concepts for safety finetuning"
)
print(str(response))

The key concepts for safety fine-tuning include supervised safety fine-tuning, safety RLHF (Reinforcement Learning from Human Feedback), and safety context distillation. Supervised safety fine-tuning involves gathering adversarial prompts and safe demonstrations to train the model to align with safety guidelines. Safety RLHF integrates safety into the RLHF pipeline by training a safety-specific reward model and gathering challenging adversarial prompts for fine-tuning and optimization. Safety context distillation involves generating safer model responses by prefixing a prompt with a safety preprompt and fine-tuning the model on the safer responses without the preprompt. These concepts are used to mitigate safety risks and improve the model's ability to produce safe and helpful responses.

Evaluation#

We evaluate how well our recursive retrieval + node reference methods work. We evaluate both chunk references as well as metadata references. We use embedding similarity lookup to retrieve the reference nodes.

We compare both methods against a baseline retriever where we fetch the raw nodes directly.

In terms of metrics, we evaluate using both hit-rate and MRR.

Dataset Generation#

We first generate a dataset of questions from the set of text chunks.

from llama_index.evaluation import (
    generate_question_context_pairs,
    EmbeddingQAFinetuneDataset,
)
from llama_index.llms import OpenAI

import nest_asyncio

nest_asyncio.apply()

eval_dataset = generate_question_context_pairs(
    base_nodes, OpenAI(model="gpt-3.5-turbo")
)

100%|██████████| 93/93 [02:08<00:00,  1.38s/it]

eval_dataset.save_json("data/llama2_eval_dataset.json")

# optional
eval_dataset = EmbeddingQAFinetuneDataset.from_json(
    "data/llama2_eval_dataset.json"
)

Compare Results#

We run evaluations on each of the retrievers to measure hit rate and MRR.

We find that retrievers with node references (either chunk or metadata) tend to perform better than retrieving the raw chunks.

import pandas as pd
from llama_index.evaluation import RetrieverEvaluator, get_retrieval_results_df

# set vector retriever similarity top k to higher
top_k = 10


def display_results(names, results_arr):
    """Display results from evaluate."""

    hit_rates = []
    mrrs = []
    for name, eval_results in zip(names, results_arr):
        metric_dicts = []
        for eval_result in eval_results:
            metric_dict = eval_result.metric_vals_dict
            metric_dicts.append(metric_dict)
        results_df = pd.DataFrame(metric_dicts)

        hit_rate = results_df["hit_rate"].mean()
        mrr = results_df["mrr"].mean()
        hit_rates.append(hit_rate)
        mrrs.append(mrr)

    final_df = pd.DataFrame(
        {"retrievers": names, "hit_rate": hit_rates, "mrr": mrrs}
    )
    display(final_df)

vector_retriever_chunk = vector_index_chunk.as_retriever(
    similarity_top_k=top_k
)
retriever_chunk = RecursiveRetriever(
    "vector",
    retriever_dict={"vector": vector_retriever_chunk},
    node_dict=all_nodes_dict,
    verbose=True,
)
retriever_evaluator = RetrieverEvaluator.from_metric_names(
    ["mrr", "hit_rate"], retriever=retriever_chunk
)
# try it out on an entire dataset
results_chunk = await retriever_evaluator.aevaluate_dataset(
    eval_dataset, show_progress=True
)

vector_retriever_metadata = vector_index_metadata.as_retriever(
    similarity_top_k=top_k
)
retriever_metadata = RecursiveRetriever(
    "vector",
    retriever_dict={"vector": vector_retriever_metadata},
    node_dict=all_nodes_dict,
    verbose=True,
)
retriever_evaluator = RetrieverEvaluator.from_metric_names(
    ["mrr", "hit_rate"], retriever=retriever_metadata
)
# try it out on an entire dataset
results_metadata = await retriever_evaluator.aevaluate_dataset(
    eval_dataset, show_progress=True
)

base_retriever = base_index.as_retriever(similarity_top_k=top_k)
retriever_evaluator = RetrieverEvaluator.from_metric_names(
    ["mrr", "hit_rate"], retriever=base_retriever
)
# try it out on an entire dataset
results_base = await retriever_evaluator.aevaluate_dataset(
    eval_dataset, show_progress=True
)

100%|██████████| 194/194 [00:09<00:00, 19.86it/s]

full_results_df = get_retrieval_results_df(
    [
        "Base Retriever",
        "Retriever (Chunk References)",
        "Retriever (Metadata References)",
    ],
    [results_base, results_chunk, results_metadata],
)
display(full_results_df)

	retrievers	hit_rate	mrr
0	Base Retriever	0.778351	0.563103
1	Retriever (Chunk References)	0.896907	0.691114
2	Retriever (Metadata References)	0.891753	0.718440