configs/hotpotqa.yaml


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

hopfield:
  beta: 2.0
  max_iter: 8
  conv_threshold: 1.0e-4
  top_k: 5

memory:
  embedding_dim: 768
  normalize: true

encoder:
  model_name: "facebook/contriever-msmarco"
  max_length: 512
  batch_size: 64

generator:
  model_name: "meta-llama/Llama-3.1-8B-Instruct"
  max_new_tokens: 128
  temperature: 0.0

retriever_type: "hopfield"
dataset: "hotpotqa"