# yandex/YaLM-100B

Pretrained language model with 100B parameters

Repository: https://github.com/yandex/YaLM-100B
Canonical: https://ross.abutalabs.com/products/yalm-100b
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2023-07-10T00:54:27+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1533, "days_push": 1151, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3757, forks 290 (observed 2026-08-28T04:08:17.828721+00:00)

## What it is
YaLM-100B is a GPT-like pretrained language model with 100 billion parameters, trained by Yandex on English and Russian text using DeepSpeed and Megatron-LM-style infrastructure. The repository provides inference code, download scripts, and Docker tooling for running the model on multi-GPU setups with roughly 200GB of GPU memory.

## Use cases
- generate text with a large open language model
- run a 100B parameter GPT model on multiple GPUs
- do few-shot text generation in English and Russian
- download and deploy pretrained LLM weights locally
- experiment with conditional sampling and greedy decoding
- study training practices for large language models

## When to choose
- you need a large open-weight GPT-style model, especially with strong Russian-language capability
- you have a multi-GPU cluster (~200GB VRAM) and want to run inference locally
- you want to study or reproduce large-scale LLM training with DeepSpeed

## When to avoid
- you lack access to high-end GPUs or 200GB of disk space
- you need a small, lightweight model for CPU or single-GPU inference
- you need actively maintained code - the repo is a stock DeepSpeed example with minimal changes and updates have stalled

## Facets
- artifact type: library
- maturity: maintenance
- function: llm-inference, llm-training, machine-learning, deep-learning
- domain: large-language-models, deep-learning, artificial-intelligence
- platform: python
- tags: gpt-like, 100b-parameters, multilingual, english-russian, deepspeed, megatron-lm, tensor-parallelism, text-generation, pretrained-model, yandex, natural-language-processing, gpu, docker, linux

## Member repositories
- yandex/YaLM-100B (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:17.828721+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 2026-08-29T18:28:56.350874+00:00, confidence not recorded.
  - readme: https://github.com/yandex/YaLM-100B (fetched 2026-08-28T04:08:17.828721+00:00, sha 7274e612b5ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.