nlp-thirdgate

A hub of third-party NLP providers and tutorials to help you instantly handle your data iterator with no-string dependency apps.

The purpose is of this project is to share Third-party providers that could be combined into a single pipeline.

Third-Party Providers

• LLM / Mistral.AI [🤖 models]
• LLM / OpenRouter.AI [🤖 models]
• LLM / Replicate.IO [🤖 models]
• LLM / OpenAI:
  • ChatGPT
  • Qwen-2.5-Max
  • o1
• LLM / Transformers:
  • DeepSeek-R1-distill-7b [📙 qwen-notebook] [📙 llama3-notebook]
  • LLaMA-3
  • Qwen-2
  • Phi-4
  • Gemma-3 [📙 notebook]
  • Flan-T5
  • Mistral
• NER / DeepPavlov [📙 notebook]
• NER / Flair [bash-script] [🤖 models]
• NER / Spacy [bash-script] [🤖 models]
• Translation / GoogleTranslator [📙 notebook]

Data Iterators

In this project we consider that each provider represent a wrapper over third-party app to handle iterator of data. We consider dict python type for representing each record of the data.

Pipeline Formation

If you wish to use several third-party providers all together for a data-iterators, it is recommented to adopt AREkit framework as a no-string solution for deploying pipeline that support batching mode.

No-string Application

• bulk-chain -- framework for reasoning over your tabular data rows with any provided LLM
• bulk-ner -- framework for a quick third-party models binding for entities extraction from cells of long tabular data
• bulk-translate -- framework for translation of a massive stream of texts with native support of pre-annotated fixed-spans that are invariant for translator.
• AREkit pipelines -- toolkit for handling your textual data iterators with various NLP providers