# bilibili/Index-1.9B

A lightweight multilingual LLM

Repository: https://github.com/bilibili/Index-1.9B
Canonical: https://ross.abutalabs.com/products/index-19b
Language: Python
License: Apache-2.0
License Family: permissive
Topics: llm, nlp
Last push: 2026-08-21T13:13:47+00:00

## Health v2 (maintenance only)
Score: 68/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 35, longevity 58
- inputs: {"age_days": 813, "days_push": 12, "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 1030, forks 51 (observed 2026-08-28T04:03:17.904878+00:00)

## What it is
Index-1.9B is a family of lightweight 1.9-billion-parameter multilingual language models from Bilibili's Index team, released in base, chat, character (few-shot role-playing), and 32K long-context variants. The Apache-2.0 weights are hosted on HuggingFace with GGUF builds adapted for llama.cpp and Ollama, pre-trained on roughly 2.8T tokens of mostly Chinese and English corpus.

## Use cases
- run a small multilingual chatbot locally on consumer hardware
- translate between Chinese and other East Asian languages with a lightweight model
- build a role-playing character chatbot with few-shot customization
- read and summarize long documents with a tiny 32K-context model
- study a pre-trained base model and constant-LR checkpoints for training research
- embed an offline LLM into an edge or low-resource application

## When to choose
- You need a sub-2B-parameter model that runs on modest GPUs or CPUs via GGUF/Ollama
- Your work centers on Chinese and English text, including East Asian translation
- You want character role-playing chat without heavy fine-tuning (few-shot plus RAG)
- You need surprisingly long 32K context from a very small model

## When to avoid
- You need frontier-grade reasoning, coding, or agentic capabilities from a larger model
- You require broad multilingual coverage beyond Chinese/English and East Asian languages
- Your tasks demand context lengths well beyond 32K tokens

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, nlp, chatbot
- domain: artificial-intelligence, large-language-models
- platform: python, cross-platform, self-hosted
- tags: llm, small-language-model, pretrained-model, model-weights, multilingual, chinese-english, chat-model, role-playing, few-shot-roleplay, rag, long-context, 32k-context, gguf, ollama, llama-cpp, sft, dpo, huggingface, bilibili, natural-language-processing

## Member repositories
- bilibili/Index-1.9B (main) score 68

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:17.904878+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 2026-08-30T07:07:32.616019+00:00, confidence not recorded.
  - readme: https://github.com/bilibili/Index-1.9B (fetched 2026-08-28T04:03:17.904878+00:00, sha 9028c784b22e)
- Data as of 2026-08-30T08:39:29.467469+00:00.