# salesforce/awd-lstm-lm

LSTM and QRNN Language Model Toolkit for PyTorch

Repository: https://github.com/salesforce/awd-lstm-lm
Canonical: https://ross.abutalabs.com/products/awd-lstm-lm
Language: Python
License: BSD-3-Clause
License Family: permissive
Topics: lstm, pytorch, language-model, sgd, qrnn
Archived: true
Last push: 2022-02-12T14:33:58+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3312, "days_push": 1663, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1989, forks 484 (observed 2026-08-28T04:06:02.987543+00:00)

## What it is
A PyTorch toolkit for training LSTM and QRNN language models, implementing Salesforce Research's AWD-LSTM papers. It supports word-level and character-level language modeling on standard datasets like Penn Treebank and WikiText.

## Use cases
- train a word-level language model on WikiText-2
- reproduce AWD-LSTM results from the paper
- train a character-level language model on enwik8
- compare LSTM vs QRNN language model performance
- apply continuous cache pointer to a finetuned language model
- learn regularized LSTM language modeling in PyTorch

## When to choose
- you need to reproduce or build on the AWD-LSTM/QRNN language model papers
- you want a proven baseline for word or character-level language modeling on PTB or WikiText
- you want a faster QRNN alternative to cuDNN LSTM

## When to avoid
- you need transformer-based language models or modern LLM tooling
- you require support for recent PyTorch versions (code targets PyTorch 0.4)
- you want a maintained production library rather than research code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, nlp
- domain: deep-learning, machine-learning
- platform: python
- tags: lstm, qrnn, language-model, pytorch, research-code, awd-lstm, natural-language-processing, gpu

## Member repositories
- salesforce/awd-lstm-lm (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:02.987543+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 2026-08-30T03:02:57.002977+00:00, confidence not recorded.
  - readme: https://github.com/salesforce/awd-lstm-lm (fetched 2026-08-28T04:06:02.987543+00:00, sha 25aa660fb561)
- Data as of 2026-08-30T08:39:29.467469+00:00.