# NVIDIA/sentiment-discovery

Unsupervised Language Modeling at scale for robust sentiment classification

Repository: https://github.com/NVIDIA/sentiment-discovery
Canonical: https://ross.abutalabs.com/products/sentiment-discovery
Language: Python
License: NOASSERTION
License Family: other
Last push: 2020-06-28T23:18:01+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3198, "days_push": 2257, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1065, forks 208 (observed 2026-08-28T04:03:26.680283+00:00)

## What it is
A deprecated PyTorch codebase from NVIDIA for large-scale unsupervised language model pretraining and transfer to sentiment and emotion classification tasks. It supports mixed-precision and distributed multi-GPU training, and has been superseded by NVIDIA's Megatron-LM.

## Use cases
- train a transformer language model on large text corpora
- classify sentiment of product reviews
- finetune pretrained language models on custom text classification datasets
- reproduce OpenAI's sentiment neuron paper at scale
- classify tweets into emotion categories
- featurize text for few-shot classification tasks

## When to choose
- you need the exact code from NVIDIA's 2018-2019 large-scale language modeling papers
- you want to reproduce the unsupervised sentiment discovery results with tagged historical releases

## When to avoid
- you want actively maintained large language model training code - use Megatron-LM instead
- you need support for recent PyTorch versions or modern hardware
- you are starting a new sentiment analysis project

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, nlp, llm-training, cli
- domain: machine-learning, deep-learning
- platform: python
- tags: sentiment-classification, transfer-learning, language-modeling, pytorch, deprecated, multi-gpu, mixed-precision, natural-language-processing, linux, gpu

## Member repositories
- NVIDIA/sentiment-discovery (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:26.680283+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:55:28.001154+00:00, confidence not recorded.
  - readme: https://github.com/NVIDIA/sentiment-discovery (fetched 2026-08-28T04:03:26.680283+00:00, sha 827a8f52aea7)
- Data as of 2026-08-30T08:39:29.467469+00:00.