codelion/hash-hop Long Context Evaluation

[ShellLM]

• hash-hop/README.md at main · codelion/hash-hop

HashHop Long Context Evaluation

This repository contains the code for HashHop, our long context architecture benchmark.

Installation Guide

Prerequisites

• Git
• Python 3.9+
• Poetry

Steps

1. Clone the repository:

   git clone git@github.com:magicproduct/hash-hop.git
   cd hash-hop
   

2. Install dependencies:

   poetry install
   

Generating Evaluation Data

The MultiHopEval.make_one function generates a MultiHopSample object which can be used for either evaluation (via the targets field) or for training models on the multihop task (via the completion field).

Usage Example

from hashhop import MultiHopEval

CHARS_PER_TOKEN = 3
datapoint = MultiHopEval.make_one(
    n_chars_problem=int(1_000_000 * CHARS_PER_TOKEN),
    num_queries=5,
    hops=2,
    hash_pair_str_length=16,
    chain_of_thought=False,
)
print(datapoint.prompt)
print(datapoint.completion)
print(datapoint.targets)

Parameters

• n_chars_problem: int
  • The size of the problem in characters.
• num_queries: int
  • The number of queries in the completion.
• hops: int
  • The number of hops in the reasoning chain.
• hash_pair_str_length: int
  • The number of characters per hash.
• chain_of_thought: bool
  • If True, the model is asked to produce H1 -> H2 -> H3.
  • If False, the model is asked to produce H1 -> H3.

Output

• prompt: str
  • Contains the shuffled hash pairs.
• (Used for training) completion: str
  • The queries and targets in string format
• (Used for evaluation) targets: Dict[str, str]
  • Contains query-ground truth pairs in structured format
  • If chain of thought is false, will contain {H1: H3} (e.g. 'HETyxiWTFSVUYega': 'pChfybAJRUBmdAGC')
  • If chain of thought is true, will contain full chain {H1: H2 = H3} (e.g. 'KeiVcwXpnYIWLPmk': 'GmmNmICdvEErHgei = JhgvBFdYCnLVZBoy')

Citation

@misc{magic2024hashhop,
  author = {Magic},
  title = {HashHop: Long Context Evaluation},
  year = {2024},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\\url{https://github.com/magicproduct/hash-hop}},
}

License

MIT

Suggested labels

None

[ShellLM]

Related content

#309 similarity score: 0.89
#919 similarity score: 0.89
#848 similarity score: 0.89
#456 similarity score: 0.88
#811 similarity score: 0.88
#324 similarity score: 0.88