Skip to content

Train

JSONL Format

The text field contains the text sample for the training case. The label fields must be an int to represent what ID the text belongs to.

{"text": "Sample train case ", "label": 0} 
{"text": "Another sample train case ", "label": 1}

Labels IDs

You can see what labels and IDs are available for your model with the following code

from erictransformer import EricTextClassification

eric_tc_labels = EricTextClassification(model_name="bert-base-uncased", labels=["custom_0", "custom_1", "custom_3"])

print(eric_tc_labels.config.id2label)  # {0: 'custom_0', 1: 'custom_1', 2: 'custom_3'}
print(eric_tc_labels.config.label2id)  # {'custom_0': 0, 'custom_1': 1, 'custom_3': 2}

# when fine-tuning provide the ID 0 for custom_0, 1 for 'custom_1' and 2 for 'custom_3'

train()

inputs: 1. train_path (string) (required): a path file to a train file or a directory that contains train files.

  1. eval_filepath (string) (optional): a path file to an eval file or a directory that contains eval files.

  2. args (EricTrainArgs) (optional): a dataclass with the arguments found here.

import json
from erictransformer import EricTextClassification, EricTrainArgs

eric_gen = EricTextClassification(model_name="bert-base-uncased", labels=["LABEL_0", "LABEL_1", "LABEL_3"])
args = EricTrainArgs(out_dir="eric_transformer", lr=2e-5)
train_data = [{"text": "Train data 0", "label": 0},
              {"text": "Train data 1", "label": 1},
              {"text": "Train data 2", "label": 2}
              ]

with open("data.jsonl", "w") as f:
    for td in train_data:
        f.write(json.dumps(td) + "\n")

result = eric_gen.train(train_path="data.jsonl", eval_path="data.jsonl", args=args)

print(result)

View the output directory in eric_transformer/

eval()

Inputs:

  1. train_path (string) (required): Same as train()'s train_path parameter.

  2. args (EricEvalArgs) (optional): a dataclass with these arguments.

import json
from erictransformer import EricTextClassification, EricEvalArgs

eric_tc = EricTextClassification(model_name="bert-base-uncased")
args = EricEvalArgs(out_dir="eric_transformer")

train_data = [{"text": "Eval data 0", "label": 0}, {"text": "Eval data 1", "label": 1},
              ]
with open("eval.jsonl", "w") as f:
    for td in train_data:
        f.write(json.dumps(td) + "\n")

result = eric_tc.eval("eval.jsonl", args=args)

print(result.loss)

You can view the tokenized data in eric_transformer/