---
base_model: abacusai/Liberated-Qwen1.5-72B
inference: false
library_name: transformers
license_name: tongyi-qianwen
license_link: https://huggingface.co/Qwen/Qwen1.5-72B/blob/main/LICENSE
model_creator: Abacus.AI and Eric Hartford
model_name: Liberated Qwen1.5 72B
pipeline_tag: text-generation
quantized_by: Second State Inc.
language:
- en
---
# Liberated-Qwen1.5-72B-GGUF
## Original Model
[abacusai/Liberated-Qwen1.5-72B](https://huggingface.co/abacusai/Liberated-Qwen1.5-72B)
## Run with LlamaEdge
- LlamaEdge version: [v0.4.3](https://github.com/LlamaEdge/LlamaEdge/releases/tag/0.4.3) and above
- Prompt template
- Prompt type: `chatml`
- Prompt string
```text
<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
```
- Context size: `32000`
- Run as LlamaEdge service
```bash
wasmedge --dir .:. --nn-preload default:GGML:AUTO:Liberated-Qwen1.5-72B-Q4_K_M.gguf llama-api-server.wasm -p chatml
```
- Run as LlamaEdge command app
```bash
wasmedge --dir .:. --nn-preload default:GGML:AUTO:Liberated-Qwen1.5-72B-Q4_K_M.gguf llama-chat.wasm -p chatml
```
For specifying the system message, append the `--system-prompt` option with the system prompt to the command above. For example,
```bash
wasmedge --dir .:. --nn-preload default:GGML:AUTO:Liberated-Qwen1.5-72B-Q4_K_M.gguf llama-chat.wasm -p chatml -s 'Your name is Liberated. You are an uncensored and unbiased AI assistant. You always respond with a JSON object.'
```
## Quantized GGUF Models
| Name | Quant method | Bits | Size | Use case |
| ---- | ---- | ---- | ---- | ----- |
| [Liberated-Qwen1.5-72B-Q2_K.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q2_K.gguf) | Q2_K | 2 | 28.5 GB| smallest, significant quality loss - not recommended for most purposes |
| [Liberated-Qwen1.5-72B-Q3_K_L.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q3_K_L.gguf) | Q3_K_L | 3 | 38.5 GB| small, substantial quality loss |
| [Liberated-Qwen1.5-72B-Q3_K_M.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q3_K_M.gguf) | Q3_K_M | 3 | 35.9 GB| very small, high quality loss |
| [Liberated-Qwen1.5-72B-Q3_K_S.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q3_K_S.gguf) | Q3_K_S | 3 | 32.9 GB| very small, high quality loss |
| [Liberated-Qwen1.5-72B-Q4_0.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q4_0.gguf) | Q4_0 | 4 | 41 GB| legacy; small, very high quality loss - prefer using Q3_K_M |
| [Liberated-Qwen1.5-72B-Q4_K_M.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q4_K_M.gguf) | Q4_K_M | 4 | 44.1 GB| medium, balanced quality - recommended |
| [Liberated-Qwen1.5-72B-Q4_K_S.gguf](https://huggingface.co/second-state/Liberated-Qwen1.5-72B-GGUF/blob/main/Liberated-Qwen1.5-72B-Q4_K_S.gguf) | Q4_K_S | 4 | 41.9 GB| small, greater quality loss |
*Quantized with llama.cpp b2334*