# young-geng/EasyLM

Large language models (LLMs) made easy, EasyLM is a one stop solution for pre-training, finetuning, evaluating and serving LLMs in JAX/Flax.

Repository: https://github.com/young-geng/EasyLM
Canonical: https://ross.abutalabs.com/products/easylm
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, flax, jax, language-model, natural-language-processing, transformer, large-language-models, chatbot, llama
Last push: 2024-08-13T05:55:05+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 98
- inputs: {"age_days": 1380, "days_push": 750, "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 2514, forks 259 (observed 2026-08-28T04:06:57.588625+00:00)

## What it is
EasyLM is a JAX/Flax-based framework for pre-training, finetuning, evaluating, and serving large language models like LLaMA. It scales training across hundreds of TPU/GPU accelerators using JAX's pjit model sharding.

## Use cases
- pretrain large language models on tpu pods
- finetune llama on multiple gpus
- train llms in jax with model sharding
- serve a finetuned llama chatbot locally
- reproduce llama with permissive license weights
- multi-host distributed llm training

## When to choose
- you want to train or finetune LLaMA-family models in JAX/Flax
- your model doesn't fit on a single accelerator and needs pjit sharding
- you're running on Google Cloud TPU Pods or multi-GPU hosts
- you want a simpler, customizable codebase than heavier training frameworks

## When to avoid
- you prefer PyTorch-based training ecosystems
- you only need inference of existing models without training
- you need broad model support beyond the LLaMA family
- you need a framework with frequent updates and active maintenance

## Facets
- artifact type: library
- maturity: maintenance
- function: llm-training, llm-inference, deep-learning, machine-learning
- domain: large-language-models, deep-learning, machine-learning
- platform: python, cloud
- tags: jax, flax, llama, tpu, model-sharding, pjit, finetuning, huggingface, natural-language-processing, gpu, linux

## Member repositories
- young-geng/EasyLM (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:57.588625+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 2026-08-30T02:26:18.529420+00:00, confidence not recorded.
  - readme: https://github.com/young-geng/EasyLM (fetched 2026-08-28T04:06:57.588625+00:00, sha b42b9c51bfb9)
- Data as of 2026-08-30T08:39:29.467469+00:00.