[whatmodel.sh](../../README.md) / Qwen2-Audio 7B Instruct
# Qwen2-Audio 7B Instruct
Alibaba · released 2024-07 · 8.4B params · apache-2.0
**score 44.1** ±2.5 · #499 of 732 · from 10 benchmark results across 1 source
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen2-Audio 7B Instruct among every model on that task or test, worst to best.
other tests (10)
test result vs the others in its run difficulty counts why
AHELM Speech Robust Bench (LibriS… 39.9% ▂·●··▆▄·▂▂·▄ beats 4/13 avg 34% 45% typical
AHELM CoRe-Bench conversational a… 23.3% ●·▄··▂▂·▄▂█▄ beats 0/13 medium · avg 52% 22% at its level
AHELM LibriSpeech transcription b… 11.3% ▃···▃·●▃▅▃▅█ beats 2/13 avg 11% 11% typical
AHELM MUStARD audio sarcasm detec… 20.9% ●·······▆█▆▆ beats 0/13 very hard · avg 57% 5.4% blunt
AHELM AIR-Bench Foundation music … 72.4% ▃█▃▅▃·▃·▃▅●▃ beats 12/13 medium · avg 57% 4.3% blunt · at its level
AHELM VoxCeleb2 same-speaker dete… 24.0% ●·▂·▂█▂▄▄··▄ beats 0/13 very hard · avg 54% 3.5% sharp · too hard for it · noisy
AHELM MELD audio emotion classifi… 26.0% ●··▃▃▃·▃▅▅█▅ beats 0/13 medium · avg 39% 3.2% blunt · at its level
AHELM FLEURS 7-language transcrip… 224.0% ●▂···▂▆▂█▂·▂ beats 1/13 avg 83% 2.2% blunt · noisy
AHELM Multilingual LibriSpeech 7-… 33.7% ▂▂●··▂▂·▂▂█▆ beats 2/13 avg 14% 0.4% blunt · noisy
AHELM VocalSound human-sound reco… 79.9% ██▄▄·█·●·█▄▄ beats 9/13 easy · avg 67% 0.0% blunt · at its level · noisy
source results share of the score
AHELM: A Holistic Evaluation of Audio-L… 10 97% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle 3% ▇
weighted average = 44.1
Licences: AHELM: A Holistic Evaluation of Audio-Language Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Qwen2-Audio 7B Instruct (Q4_K_M, needs ~4.8 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF/tree/e1e68850ba33e38eafbc3817919c318d9c7e757b))
```sh
llama cli -hf mradermacher/Qwen2-Audio-7B-Instruct-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf mradermacher/Qwen2-Audio-7B-Instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~8.2 GB) `llama cli -hf mradermacher/Qwen2-Audio-7B-Instruct-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Qwen2-Audio 7B Instruct (Q4_K_M, needs ~4.8 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF/tree/e1e68850ba33e38eafbc3817919c318d9c7e757b))
```sh
lms get https://huggingface.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~8.2 GB) `lms get https://huggingface.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF@Q8_0 -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Qwen2-Audio 7B Instruct (Q4_K_M, needs ~4.8 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF/tree/e1e68850ba33e38eafbc3817919c318d9c7e757b))
```sh
ollama run hf.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~8.2 GB) `ollama run hf.co/mradermacher/Qwen2-Audio-7B-Instruct-GGUF:Q8_0`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-Audio-7B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen2-audio-7b-instruct) · [JSON](https://whatmodel.sh/api?format=json&all=1)