当我运行命令时,ant得到了这个问题。allennlp的版本为0.9。
allennlp train /root/code/python/share/scidocs/data/recomm/train_similar_papers_model.json -s /root/code/python/share/scidocs/data/recomm-tmp --include-package scidocs.recommender
train_similar_papers_model.json的关键信息是:
local stringToBool(s) =
if s == "true" then true
else if s == "false" then false
else error "invalid boolean: " + std.manifestJson(s);
local EMBEDDINGS_PATH = std.extVar("EMBEDDINGS_PATH");
local CUDA_DEVICE = std.parseInt(std.extVar("CUDA_DEVICE"));
local EMBEDDINGS_DIM = std.parseInt(std.extVar("EMBEDDINGS_DIM"));
local PAPER_METADATA_PATH = std.extVar("PAPER_METADATA_PATH");
local TRAIN_PATH = std.extVar("TRAIN_PATH");
local VALID_PATH = (if std.extVar("VALID_PATH")!="" then std.extVar("VALID_PATH") else null);
local TEST_PATH = std.extVar("TEST_PATH");
local PROP_SCORE_PATH = std.extVar("PROP_SCORE_PATH");
{
"random_seed": 7,
"pytorch_seed": 7,
"numpy_seed": 7,
"dataset_reader": {
"type": "simclick_data_reader",
"paper_features_path": PAPER_METADATA_PATH,
"paper_embeddings_path": EMBEDDINGS_PATH,
"max_results_per_query": 10,
"jsonlines_embedding_format": stringToBool(std.extVar('jsonlines_embedding_format'))
},
....
}发布于 2022-03-01 01:39:57
您的配置包含:"jsonlines_embedding_format":stringToBool(std.extVar('jsonlines_embedding_format'))}
它试图读取名为jsonlines_embedding_format的环境变量,并将其转换为布尔值。
错误说明没有定义环境变量,因此在调用train命令之前,您需要首先定义它。
https://stackoverflow.com/questions/71289974
复制相似问题