Tables Packs

Chain-of-table Pack

This LlamaPack implements the Chain-of-Table paper by Wang et al..

Chain-of-Table proposes the following: given a user query over tabular data, plan out a sequence of tabular operations over the table to retrieve the right information in order to satisfy the user query. The updated table is explicitly used/modified throughout the intermediate chain (unlike chain-of-thought/ReAct which uses generic thoughts).

There is a fixed set of tabular operations that are defined in the paper:

• f_add_column
• f_select_row
• f_select_column
• f_group_by
• f_sort_by

We implemented the paper based on the prompts described in the paper, and adapted it to get it working. That said, this is marked as beta, so there may still be kinks to work through. Do you have suggestions / contributions on how to improve the robustness? Let us know!

A full notebook guide can be found here.

CLI Usage

You can download llamapacks directly using llamaindex-cli, which comes installed with the llama-index python package:

llamaindex-cli download-llamapack ChainOfTablePack --download-dir ./chain_of_table_pack

You can then inspect the files at ./chain_of_table_pack and use them as a template for your own project!

Code Usage

We will show you how to import the agent from these files!

from llama_index.core.llama_pack import download_llama_pack

# download and install dependencies
ChainOfTablePack = download_llama_pack(
    "ChainOfTablePack", "./chain_of_table_pack"
)

From here, you can use the pack. You can import the relevant modules from the download folder (in the example below we assume it's a relative import or the directory has been added to your system path).

from chain_of_table_pack.base import ChainOfTableQueryEngine, serialize_table

query_engine = ChainOfTableQueryEngine(df, llm=llm, verbose=True)
response = query_engine.query(
    "Who won best Director in the 1972 Academy Awards?"
)

You can also use/initialize the pack directly.

from llm_compiler_agent_pack.base import ChainOfTablePack

agent_pack = ChainOfTablePack(df, llm=llm, verbose=True)

The run() function is a light wrapper around agent.chat().

response = pack.run("Who won best Director in the 1972 Academy Awards?")

Mix-Self-Consistency Pack

This LlamaPack implements the mix self-consistency method proposed in "Rethinking Tabular Data Understanding with Large Language Models" paper by Liu et al.

LLMs can reason over tabular data in 2 main ways:

1. textual reasoning via direct prompting
2. symbolic reasoning via program synthesis (e.g. python, SQL, etc)

The key insight of the paper is that different reasoning pathways work well in different tasks. By aggregating results from both with a self-consistency mechanism (i.e. majority voting), it achieves SoTA performance.

We implemented the paper based on the prompts described in the paper, and adapted it to get it working. That said, this is marked as beta, so there may still be kinks to work through. Do you have suggestions / contributions on how to improve the robustness? Let us know!

A full notebook guide can be found here.

CLI Usage

You can download llamapacks directly using llamaindex-cli, which comes installed with the llama-index python package:

llamaindex-cli download-llamapack MixSelfConsistencyPack --download-dir ./mix_self_consistency_pack

You can then inspect the files at ./mix_self_consistency_pack and use them as a template for your own project!

Code Usage

We will show you how to import the module from these files!

from llama_index.core.llama_pack import download_llama_pack

# download and install dependencies
MixSelfConsistencyPack = download_llama_pack(
    "MixSelfConsistencyPack", "./mix_self_consistency_pack"
)

From here, you can use the pack. You can import the relevant modules from the download folder (in the example below we assume it's a relative import or the directory has been added to your system path).

from mix_self_consistency_pack.base import MixSelfConsistencyQueryEngine

query_engine = MixSelfConsistencyQueryEngine(df=df, llm=llm, verbose=True)
response = query_engine.query(
    "Who won best Director in the 1972 Academy Awards?"
)

You can also use/initialize the pack directly.

from mix_self_consistency_pack.base import MixSelfConsistencyPack

pack = MixSelfConsistencyPack(df=df, llm=llm, verbose=True)

The run() function is a light wrapper around query_engine.query().

response = pack.run("Who won best Director in the 1972 Academy Awards?")