[whatmodel.sh](../../README.md) / Pythia 160M
# Pythia 160M
EleutherAI · released 2023-02 · 0.162B params · apache-2.0
**score -2.1** ±1.7 · #726 of 732 · from 13 benchmark results across 4 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Pythia 160M among every model on that task or test, worst to best.
other tests (11)
test result vs the others in its run difficulty counts why
SciQ (maker) 74.1% ▂··●▃▂▂▆▃▃█▂ beats 1/19 very easy · avg 98% 22% blunt · at its level
ARC Easy (maker) 43.5% ▃·●▃▃▅·██▃██ beats 2/19 easy · avg 87% 21% blunt · at its level · several runs
ImageNetVC visual commonsense, te… 34.1% ▂·●·▂█▄▄█▄▂▆ beats 2/21 easy · avg 74% 11% blunt
PIQA (maker) 62.7% ▂·●▄·▄▂▆▂▄█▄ beats 1/19 easy · avg 87% 11% blunt · at its level
LAMBADA (OpenAI version), last-wo… 32.8% ▃··●▃▃▅▅▅███ beats 1/19 very easy · avg 87% 10% blunt · at its level
WinoGrande (maker) 53.1% ▄····▄●▂▄▆▆█ beats 4/19 easy · avg 87% 5.6% too hard for it
ARC Challenge (maker) 18.8% ●··▁·▁▁·▁▂▁▂ beats 8/19 easy · avg 74% 5.0% sharp · too hard for it · small run · s…
HellaSwag (maker) 29.4% ●▁▁▂▂▂▄▆█▇▄▂ beats 3/135 easy · avg 83% 4.9% small run
TriviaQA (maker) 1.1% ●▁▂▂▄▄▃▆▅█▅▂ beats 0/80 medium · avg 61% 3.7% too hard for it · small run
OpenBookQA (maker) 17.2% ●▁▁▁▂▂▃█▃▁▃▁ beats 1/107 very hard · avg 29% 3.1% too hard for it · small run
Open PT-LLM Leaderboard average (… 29.4% ●▁▁▁▁▁▁▁▂█▅▁ beats 10/276 medium · avg 72% 0.5% too hard for it · noisy
source results share of the score
Pythia: A Suite for Analyzing Large Lan… 6 68% ▇▇▇▇▇▇▇▇
trlX: A Framework for Large Scale Reinf… 5 19% ▇▇
ImageNetVC: Zero- and Few-Shot Visual C… 1 11% ▇
Open Portuguese LLM Leaderboard 1 0% ▇
prior: pull toward the middle 1% ▇
weighted average = -2.1
Licences: Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling ([CC BY-SA 4.0](../../licenses/paper-2304.01373v2-CC-BY-SA-4.0.txt)) · trlX: A Framework for Large Scale Reinforcement Learning from Human Feedback ([CC BY 4.0](../../licenses/acl-2023-emnlp-main-530-CC-BY-4-0.txt)) · ImageNetVC: Zero- and Few-Shot Visual Commonsense Evaluation on 1000 ImageNet Categories ([CC BY 4.0](../../licenses/acl-2023-findings-emnlp-133-CC-BY-4-0.txt)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-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 Pythia 160M (Q4_K_M, needs ~0.1 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/pythia-160m-GGUF/tree/7a79d8f693376ce9f1ba21e2c0586c3bc874e68c))
```sh
llama cli -hf mradermacher/pythia-160m-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf mradermacher/pythia-160m-GGUF:Q4_K_M
```
Other versions: Q8_0 (~0.2 GB) `llama cli -hf mradermacher/pythia-160m-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 Pythia 160M (Q4_K_M, needs ~0.1 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/pythia-160m-GGUF/tree/7a79d8f693376ce9f1ba21e2c0586c3bc874e68c))
```sh
lms get https://huggingface.co/mradermacher/pythia-160m-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 (~0.2 GB) `lms get https://huggingface.co/mradermacher/pythia-160m-GGUF@Q8_0 -y` · fp16 (~0.3 GB, Mac only) `lms get https://huggingface.co/masahiroid/pythia-160m-mlx -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 Pythia 160M (Q4_K_M, needs ~0.1 GB, from [huggingface.co/mradermacher](https://huggingface.co/mradermacher/pythia-160m-GGUF/tree/7a79d8f693376ce9f1ba21e2c0586c3bc874e68c))
```sh
ollama run hf.co/mradermacher/pythia-160m-GGUF:Q4_K_M
```
Other versions: Q8_0 (~0.2 GB) `ollama run hf.co/mradermacher/pythia-160m-GGUF: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 Pythia 160M (fp16, needs ~0.3 GB, Mac only, from [huggingface.co/masahiroid](https://huggingface.co/masahiroid/pythia-160m-mlx/tree/423ac1eacd63a9b421143288136aaaf1f838fb74))
```sh
mlx_lm.chat --model masahiroid/pythia-160m-mlx
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model masahiroid/pythia-160m-mlx
```
### Other ways to run it
1. Run it as published (fp16, needs ~0.37 GB, from [huggingface.co/EleutherAI](https://huggingface.co/EleutherAI/pythia-160m/tree/50f5173d932e8e61f858120bcb800b97af589f46))
```sh
uv run --with torch --with 'transformers<5' python -c "from transformers import pipeline; print(pipeline('text-generation', model='EleutherAI/pythia-160m')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"
```
Model repository and documentation: [huggingface.co/EleutherAI](https://huggingface.co/EleutherAI/pythia-160m)
[interactive version](https://whatmodel.sh/model/pythia-160m) · [JSON](https://whatmodel.sh/api?format=json&all=1)