unsloth / text-generation updated 10 months ago

Llama-3.2-3B-Instruct-GGUF

We have a free Google Colab Tesla T4 notebook for Llama 3.2 (3B) here: https://colab.research.google.com/drive/1T5-zKWM5OD21QHwXHiV9ixTRR7k3iB9?usp=sharing

Params
Context
131,072
Downloads 30d
224 K
Likes
80
Commercial use: conditional · llama3.2 Not gated GGUF 1 languages View on Hugging Face ↗

Download history

daily snapshots · 17 days
242 K192 K
Sep 5Sep 10Sep 16Sep 21

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

FileQuantSizeEst. VRAMVerdict on RTX 4090 · 24 GB
Llama-3.2-3B-Instruct-Q2_K.gguf Q2_K 1.4 GB 2.0 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-Q2_K_L.gguf Q2_K_L 1.4 GB 2.0 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-Q3_K_S.gguf Q3_K_S 1.5 GB 2.2 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-Q3_K_M.gguf Q3_K_M 1.7 GB 2.4 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-IQ4_XS.gguf IQ4_XS 1.8 GB 2.5 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-IQ4_NL.gguf IQ4_NL 1.9 GB 2.6 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-Q4_0.gguf Q4_0 1.9 GB 2.6 GB ✅ Runs comfortably
Llama-3.2-3B-Instruct-BF16.gguf GGUF 6.4 GB 7.6 GB ✅ Runs comfortably
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Run it

copy-paste, exact tags checked against the Hub
~ · ollama · Q4_0
$ ollama run llama-3-2-3b-instruct-gguf-unsloth

# pin the quantization explicitly
$ ollama run llama-3-2-3b-instruct-gguf-unsloth-q4_0
est. VRAM 2.6 GBon RTX 4090 · 24 GBJSON API →

Specifications

Architecture
LlamaForCausalLM
Context length
131,072
Vocabulary
128,256
Layers / heads
28 / 24
Licence
llama3.2
First seen on the Hub
2024-09-25
Base model
Llama-3.2-3B-Instruct
Training datasets
undisclosed
Added to our catalog
2026-09-05

Family

Base model and the most-downloaded derivatives in the catalog.

Compare with any text-generation model