Skip to content

Latest commit

 

History

History
59 lines (54 loc) · 6.03 KB

README.md

File metadata and controls

59 lines (54 loc) · 6.03 KB

Awesome SLMs

Description of the image

This is the list of the SLMs I use on my Raspberry Pi5 (8GB RAM) with Ollama

Name Size Tag Remark Kind URL Good on Pi5 Usable on Pi5
codegemma:2b 1.6GB 2B Fill-in-the-middle code completion code Link
gemma:2b 1.7GB 2B chat Link
gemma2:2b 1.6GB 2B chat Link
all-minilm:22m 46MB 22M Only Embeddings embedding Link
all-minilm:33m 67MB 33M Only Embeddings embedding Link
deepseek-coder:1.3b 776MB 1.3B Trained on both 87% code and 13% natural language code Link
tinyllama 638MB 1.1B chat Link
tinydolphin 637MB 1.1B chat Link
phi3:mini 2.4GB 3B chat Link
phi3.5 2.2GB 3B chat Link
granite-code:3b 2.0GB 3B code Link
qwen2.5:0.5b 398MB 0.5B chat, tools Link
qwen2.5:1.5b 986MB 1.5B chat, tools Link
qwen2.5:3b 1.9GB 3B chat, tools Link
qwen2.5-coder:1.5b 986MB 1.5B code, tools Link
qwen2:0.5b 352MB 0.5B chat Link
qwen2:1.5b 934MB 1.5B chat Link
qwen:0.5b 395MB 0.5B chat Link
qwen2-math:1.5b 935MB 1.5B Specialized math language model math Link
starcoder:1b 726MB 1B Code generation model code Link
starcoder2:3b 1.7GB 3B code Link
stablelm2:1.6b 983MB 1.6B LLM trained on multilingual data in English, Spanish, German, Italian, French, Portuguese, and Dutch. chat Link
stable-code:3b 1.6GB 3B Coding model code Link
rouge/replete-coder-qwen2-1.5b:Q8 1.9GB 1.5B Coding capabilities + non-coding data, fully cleaned and uncensored (mat+tool? to be tested) code Link
dolphin-phi:2.7b 1.6GB 2.7B uncensored chat Link
CognitiveComputations/dolphin-gemma2:2b 1.6GB 2B chat Link
allenporter/xlam:1b 873MB 1B tools Link
sam4096/qwen2tools:0.5b 352MB 0.5B tools Link
sam4096/qwen2tools:1.5b 935MB 1.5B tools Link
mxbai-embed-large:335m 670MB 335M Only Embeddings embedding Link
nomic-embed-text:v1.5 274MB 137M Only Embeddings embedding Link
yi-coder:1.5b 866MB 1.5B Code code Link
bge-m3 1.2GB 567M Only Embeddings embedding Link
reader-lm:0.5b 352MB 0.5b convert HTML to Markdown conversion Link
reader-lm:1.5b 935MB 1.5b convert HTML to Markdown conversion Link
shieldgemma:2b 1.7GB 2b evaluate the safety of text safety Link
llama-guard3:1b 1.6GB 1b evaluate the safety of text safety Link
granite3-dense:2b 1.6GB 2b chat, tools, embedding Link
granite3-moe:1b 822MB 1b chat, tools, embedding Link
llama3.2:1b 1.3GB 1b chat, tools Link
llama3.2:3b 2.0GB 3b chat, tools Link
smollm:135m 92MB 135m 🖐️ can run on a Pi 3A+ 🤪 hard to control chat Link
smollm:360m 229MB 360m 🖐️ can run well on a Pi 4 8GB chat Link
smollm:1.7b 991MB 1.7b chat Link
smollm2:135m 271MB 135m 🖐️ can run well on a Pi 4 8GB chat, tools Link
smollm2:360m 726MB 360m 🖐️ can run on a Pi 4 8GB chat, tools Link
smollm2:1.7b 1.8GB 1.7b chat, tools Link