---
title: Llasa-1B-Preserve-TextChat
canonical_url: "https://www.modelscope.cn/models/AI-ModelScope/Llasa-1B-Preserve-TextChat"
md_url: "https://www.modelscope.cn/models/AI-ModelScope/Llasa-1B-Preserve-TextChat.md"
repository: AI-ModelScope/Llasa-1B-Preserve-TextChat
last_updated: 2025-02-14
license: cc-by-nc-4.0
pipeline_tag: text-to-speech
tasks:
  - text-to-speech
model_type:
  - llama
architectures:
  - LlamaForCausalLM
base_model:
  - meta-llama/Llama-3.2-1B-Instruct
base_model_relation: finetune
parameters: 1.8B
tensor_type:
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - zh
  - en
inference_backends:
  - "deploy_task text/emb"
  - "lmdeploy 0.9.1"
  - "lmdeploy_turbomind 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 661
stars: 0
tags:
  - Text-to-Speech
---

# Llasa-1B-Preserve-TextChat

> Llasa-1B-Preserve-TextChat - AI-ModelScope 在 ModelScope 开源的模型。Update (2025-02-13): Add Llasa finetune instruction.

AI-ModelScope/Llasa-1B-Preserve-TextChat 是 ModelScope 魔搭社区上的 1.8B 参数text-to-speech模型，采用 cc-by-nc-4.0 许可，基于 meta-llama/Llama-3.2-1B-Instruct 构建，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。

- **Repository**: AI-ModelScope/Llasa-1B-Preserve-TextChat
- **License**: cc-by-nc-4.0
- **Tasks**: text-to-speech
- **Parameters**: 1.8B
- **Base model**: meta-llama/Llama-3.2-1B-Instruct
- **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: Text-to-Speech
- **Downloads**: 661
- **Stars**: 0
- **Last updated**: 2025-02-14

Source: https://www.modelscope.cn/models/AI-ModelScope/Llasa-1B-Preserve-TextChat

---

[![arXiv](https://img.shields.io/badge/arXiv-Paper-<COLOR>.svg)](https://arxiv.org/abs/2502.04128)

**Update (2025-02-13):** Add [Llasa finetune instruction](https://github.com/zhenye234/LLaSA_training/tree/main/finetune).

These models are not mentioned in the original paper, they are essentially the same as LLaSA 1B and LLaSA 3B, except they have been fine-tuned with a mixed speech and text SFT dataset, which enables the model to retain text-based conversational abilities.
 
LLaSA: Scaling Train-Time and Inference-Time Compute for LLaMA-based Speech Synthesis