> For the complete documentation index, see [llms.txt](https://blog.sunilgudivada.dev/notebook/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://blog.sunilgudivada.dev/notebook/transformers-and-llms/nlp-tasks.md).

# NLP Tasks

NLP -> Natural language Processing 

* Classification
* Multi Classification
* Generation

### **Classification** 

Input will be simple text, and output would be predict something

**Sentiment Extraction**

**Ex:** This teddy bear is so cute

Datasets : amazon, IMDB, Twitter

**Evaluation Metrics**: 

Accuracy -> % of observation that were correctly predicted

Precision -> % of predicted positive that were correct

Recall -> % of actually positive that were correct

F1 Score -> Score that is a function of precision and recall - harmonic mean

 

why do we need metrics -> to analyze the different datasets. 

Intent detection

Language detection

Topic Modelling 

## Multi Classification

 Output would be multiple predictions

Ex: Part of speech tagging 

**Named entity recognitio**n:

Identify the different named entities from the input text 

Ex: A cute teddy bear is reading 

Teddy Bear -> entity

**Datasets**: Annotated reuters news paper (CoNLL - 2003, CoNLL++)

Dependency parsing 

Constituency parsing

**Generation** 

Input as text and output also as text and dont know what is the length of the output

Machine translation

**Datasets** : WMT'14 English French, WMT'14 English German

**Evaluation Metrics**: 

BLEU -> Bi lingual evaluation under study -> quality of text translated 

ROGUE -> Quality of text generated 

Perflexity -> Quantifies how 'surprised' the model is to see some words together

Question Answering

Summarization

Text generation