metadata

base_model: meta-llama/Llama-3.2-3B-Instruct
license: llama3.2
model_creator: meta
model_name: Llama-3.2-3B-Instruct
quantized_by: Second State Inc.
language:
  - en
  - de
  - fr
  - it
  - pt
  - hi
  - es
  - th
library_name: transformers
pipeline_tag: text-generation
tags:
  - chat
  - llama
  - llama-3

Llama-3.2-3B-Instruct-GGUF

Original Model

meta-llama/Llama-3.2-3B-Instruct

Run with LlamaEdge

LlamaEdge version: v0.14.5 and above

Prompt template

Prompt type: llama-3-chat

Prompt string

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{{ system_prompt }}<|eot_id|><|start_header_id|>user<|end_header_id|>

{{ user_message_1 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{{ model_answer_1 }}<|eot_id|><|start_header_id|>user<|end_header_id|>

{{ user_message_2 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

Context size: 128000

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Llama-3.2-3B-Instruct-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template llama-3-chat \
    --ctx-size 128000 \
    --model-name Llama-3.2-3b

Run as LlamaEdge command app

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Llama-3.2-3B-Instruct-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template llama-3-chat \
  --ctx-size 128000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Llama-3.2-3B-Instruct-Q2_K.gguf	Q2_K	2	581 MB	smallest, significant quality loss - not recommended for most purposes
Llama-3.2-3B-Instruct-Q3_K_L.gguf	Q3_K_L	3	733 MB	small, substantial quality loss
Llama-3.2-3B-Instruct-Q3_K_M.gguf	Q3_K_M	3	691 MB	very small, high quality loss
Llama-3.2-3B-Instruct-Q3_K_S.gguf	Q3_K_S	3	642 MB	very small, high quality loss
Llama-3.2-3B-Instruct-Q4_0.gguf	Q4_0	4	771 MB	legacy; small, very high quality loss - prefer using Q3_K_M
Llama-3.2-3B-Instruct-Q4_K_M.gguf	Q4_K_M	4	808 MB	medium, balanced quality - recommended
Llama-3.2-3B-Instruct-Q4_K_S.gguf	Q4_K_S	4	776 MB	small, greater quality loss
Llama-3.2-3B-Instruct-Q5_0.gguf	Q5_0	5	893 MB	legacy; medium, balanced quality - prefer using Q4_K_M
Llama-3.2-3B-Instruct-Q5_K_M.gguf	Q5_K_M	5	912 MB	large, very low quality loss - recommended
Llama-3.2-3B-Instruct-Q5_K_S.gguf	Q5_K_S	5	893 MB	large, low quality loss - recommended
Llama-3.2-3B-Instruct-Q6_K.gguf	Q6_K	6	1.02 GB	very large, extremely low quality loss
Llama-3.2-3B-Instruct-Q8_0.gguf	Q8_0	8	1.32 GB	very large, extremely low quality loss - not recommended
Llama-3.2-3B-Instruct-f16.gguf	f16	16	2.48 GB

Quantized with llama.cpp b3807