File size: 1,268 Bytes
0c1ca58 7edb04f 0c1ca58 7edb04f 0c1ca58 b86b1ee 7edb04f 0c1ca58 af3b720 0c1ca58 b86b1ee af3b720 7edb04f 0c1ca58 7edb04f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 | """
domainTokenizer v0.4.0 — Building small models that understand domain tokens, not just words.
Components:
- schema: DomainSchema, FieldSpec, FieldType
- tokenizers: DomainTokenizerBuilder, per-field tokenizers
- schemas: Predefined schemas (FINANCE, ECOMMERCE, HEALTHCARE)
- models: DomainTransformerForCausalLM, PLR, JointFusion
- training: prepare_clm_dataset, pretrain_domain_model, finetune_domain_model
"""
from .schema import DomainSchema, FieldSpec, FieldType
from .tokenizers.domain_tokenizer import DomainTokenizerBuilder
from .tokenizers.field_tokenizers import (
BaseFieldTokenizer, CalendarTokenizer, CategoricalTokenizer,
DiscreteNumericalTokenizer, MagnitudeBucketTokenizer, SignTokenizer,
)
from .models.configuration import DomainTransformerConfig
from .models.modeling import DomainTransformerForCausalLM, DomainTransformerModel
from .models.plr_embeddings import PeriodicLinearReLU
from .models.joint_fusion import JointFusionModel, DCNv2
from .training.data_pipeline import prepare_clm_dataset, pack_sequences
from .training.pretrain import pretrain_domain_model
from .training.finetune_data import DomainFinetuneDataset, prepare_finetune_dataset
from .training.finetune import finetune_domain_model
__version__ = "0.4.0"
|