[whatmodel.sh](../../README.md) / Qwen2 1.5B Instruct
# Qwen2 1.5B Instruct
Alibaba · released 2024-06 · 1.54B params · apache-2.0
**score 33.5** ±0.8 · #639 of 732 · from 14 benchmark results across 7 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen2 1.5B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁●▃▆█▄▅▅▄▄▂▁ #308 of 313* 1.3% 1
coding █··········● #1 of 2* 3.9% 1
chat & writing █··········● #1 of 2* 1.3% 1
math ▁▁●▅█▆▄█▄▂▁▁ #412 of 431 45% 4
agent tasks · #308 of 313 · 1.3% of its score
Beats [QwQ-32B-Preview](qwq-32b-preview.md), [DeepSeek-Coder-V2 Lite Instruct (16B-A2.4B)](deepseek-coder-v2-lite-instruct.md) here, though they rank higher overall. Trails [Granite 4.0 350M](granite-4-0-350m.md), [Granite 4.0 H 350M](granite-4-0-h-350m.md), [Qwen2.5 0.5B Instruct](qwen2-5-0-5b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL v3 28.2% ▁▁●▁▁▃▅█▅▄▁▁ beats 7/114 medium · avg 51% 1.3% blunt · at its level
coding · #1 of 2 · 3.9% of its score
test result vs the others in its run difficulty counts why
HumanEval (maker) 37.8% ▁▁▁▁▂▂●▆█▅▂▁ beats 60/205 easy · avg 80% 3.9% sharp · at its level · small run
chat & writing · #1 of 2 · 1.3% of its score
test result vs the others in its run difficulty counts why
IFEval prompt-level strict (maker) 29.0% ▁●··▂··▂▇█▃▃ beats 1/26 medium · avg 64% 1.3% at its level · small run
math · #412 of 431 · 45% of its score
Beats [Command R (03-2024)](command-r-03-2024.md), [Llama 2 13B Chat](llama-2-13b-chat.md) here, though they rank higher overall. Trails [GPT-Neo 125M](gpt-neo-125m.md), [OPT-125M](opt-125m.md), [GPT-2 XL (1.5B)](gpt-2-xl.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
GSM8K (maker) 64.1% ●···█▄▄·▄·▄█ beats 0/8 very easy · avg 92% 21% sharp · at its level · several runs
GSM8K (ZeroEval, zero-shot) 43.4% ●▄·▁▄█▅▅▅███ beats 0/43 very easy · avg 87% 17% sharp · at its level
MATH (subset unspecified) (maker) 25.1% ●····▃·▅·▅·█ beats 0/8 easy · avg 74% 4.3% at its level
Minerva MATH (maker) 5.5% ●·····▃··█▅▅ beats 0/8 hard · avg 34% 2.7% typical
other tests (6)
test result vs the others in its run difficulty counts why
MMLU-Redux (ZeroEval, 2,778 quest… 41.1% ●·▁▁▄▄▄▅█▄▂▁ beats 0/53 easy · avg 75% 26% at its level
LiveBench 2024-07-26 10.3% ▁▁●▂▂▃▂█▆▃▅▃ beats 4/61 medium · avg 38% 12% typical
Polish EQ-Bench v2 (SpeakLeash) 15.3% ▁··●▁▂▂▆▅▆█▆ beats 1/70 medium · avg 62% 3.1% at its level
MMLU (maker) 52.4% ▁▁▁▁▁▂▂●██▆▄ beats 58/214 easy · avg 80% 3.0% at its level · small run
Open Persian LLM Leaderboard v2 a… 36.8% ▂●▂▃█▄▃▆▃▅▂▂ beats 10/81 medium · avg 57% 2.9% blunt · at its level
C-Eval (maker) 63.8% ▁···▁▁▂▁●▂▃█ beats 9/33 easy · avg 81% 2.0% at its level · small run
source results share of the score
WildEval ZeroEval published results 2 43% ▇▇▇▇▇
Qwen (submitted / maker-reported) 3 23% ▇▇▇
Qwen2-0.5B-Instruct model card (submitt… 5 15% ▇▇
LiveBench model judgments 1 12% ▇
SpeakLeash Polish EQ-Bench leaderboard 1 3% ▇
Open Persian LLM Leaderboard v2 1 3% ▇
Berkeley Function Calling Leaderboard r… 1 1% ▇
prior: pull toward the middle 0% ▇
weighted average = 33.5
Licences: WildEval ZeroEval published results ([Apache-2.0](../../licenses/zeroeval-published-results-Apache-2.0.txt)) · Qwen2-0.5B-Instruct model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · LiveBench model judgments ([Apache-2.0 (stated in the LiveBench datasheet)](https://www.apache.org/licenses/LICENSE-2.0)) · SpeakLeash Polish EQ-Bench leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Open Persian LLM Leaderboard v2 ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Berkeley Function Calling Leaderboard result archive ([Apache-2.0](../../licenses/bfcl-result-archive-v4-Apache-2.0.txt))
## 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 1.5B Instruct (Q4_K_M, needs ~1 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-1.5B-Instruct-GGUF/tree/c62434db644497c0ee545c690bb66a67eba6eb3f))
```sh
llama cli -hf Qwen/Qwen2-1.5B-Instruct-GGUF:q4_k_m
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf Qwen/Qwen2-1.5B-Instruct-GGUF:q4_k_m
```
Other versions: Q8_0 (~1.6 GB) `llama cli -hf Qwen/Qwen2-1.5B-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 1.5B Instruct (Q4_K_M, needs ~1 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-1.5B-Instruct-GGUF/tree/c62434db644497c0ee545c690bb66a67eba6eb3f))
```sh
lms get https://huggingface.co/Qwen/Qwen2-1.5B-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 (~1.6 GB) `lms get https://huggingface.co/Qwen/Qwen2-1.5B-Instruct-GGUF@q8_0 -y` · 4bit (~0.9 GB, Mac only) `lms get https://huggingface.co/mlx-community/Qwen2-1.5B-Instruct-4bit -y` · 8bit (~1.6 GB, Mac only) `lms get https://huggingface.co/mlx-community/Qwen2-1.5B-Instruct-8bit -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 1.5B Instruct (Q4_K_M, needs ~1.42 GB, from [ollama.com](https://ollama.com/library/qwen2/tags))
```sh
ollama run qwen2:1.5b-instruct-q4_K_M
```
Other versions: Q8_0 (~2.35 GB) `ollama run qwen2:1.5b-instruct-q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Qwen2 1.5B Instruct (4bit, needs ~0.9 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/Qwen2-1.5B-Instruct-4bit/tree/5dd05592c98fb2b6e67fee122eee047ca1363146))
```sh
mlx_lm.chat --model mlx-community/Qwen2-1.5B-Instruct-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/Qwen2-1.5B-Instruct-4bit
```
Other versions: 8bit (~1.6 GB, Mac only) `mlx_lm.chat --model mlx-community/Qwen2-1.5B-Instruct-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-1.5B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen2-1-5b) · [JSON](https://whatmodel.sh/api?format=json&all=1)