README.md

metadata

base_model: nvidia/Llama-3_1-Nemotron-51B-Instruct
license: other
license_name: nvidia-open-model-license
license_link: >-
  https://developer.download.nvidia.com/licenses/nvidia-open-model-license-agreement-june-2024.pdf
model_creator: tiiuae
model_name: Llama-3_1-Nemotron-51B-Instruct
quantized_by: Second State Inc.
library_name: transformers
pipeline_tag: text-generation
language:
  - en
tags:
  - nvidia
  - llama-3
  - pytorch

---

Llama-3_1-Nemotron-51B-Instruct-GGUF

Original Model

nvidia/Llama-3_1-Nemotron-51B-Instruct

Run with LlamaEdge

• LlamaEdge version: coming soon

• 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: 8192

• Run as LlamaEdge service

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Llama-3_1-Nemotron-51B-Instruct-Q5_K_M.gguf \
    llama-api-server.wasm \
    --model-name Llama-3_1-Nemotron-51B-Instruct \
    --prompt-template llama-3-chat \
    --ctx-size 8192
  

• Run as LlamaEdge command app

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:Llama-3_1-Nemotron-51B-Instruct-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template llama-3-chat \
    --ctx-size 8192
  

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Llama-3_1-Nemotron-51B-Instruct-Q2_K.gguf	Q2_K	2	19.4 GB	smallest, significant quality loss - not recommended for most purposes
Llama-3_1-Nemotron-51B-Instruct-Q3_K_L.gguf	Q3_K_L	3	27.3 GB	small, substantial quality loss
Llama-3_1-Nemotron-51B-Instruct-Q3_K_M.gguf	Q3_K_M	3	25.2 GB	very small, high quality loss
Llama-3_1-Nemotron-51B-Instruct-Q3_K_S.gguf	Q3_K_S	3	22.7 GB	very small, high quality loss
Llama-3_1-Nemotron-51B-Instruct-Q4_0.gguf	Q4_0	4	29.3 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Llama-3_1-Nemotron-51B-Instruct-Q4_K_M.gguf	Q4_K_M	4	31.0 GB	medium, balanced quality - recommended
Llama-3_1-Nemotron-51B-Instruct-Q4_K_S.gguf	Q4_K_S	4	29.5 GB	small, greater quality loss
Llama-3_1-Nemotron-51B-Instruct-Q5_0.gguf	Q5_0	5	35.6 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Llama-3_1-Nemotron-51B-Instruct-Q5_K_M.gguf	Q5_K_M	5	36.5 GB	large, very low quality loss - recommended
Llama-3_1-Nemotron-51B-Instruct-Q5_K_S.gguf	Q5_K_S	5	35.6 GB	large, low quality loss - recommended
Llama-3_1-Nemotron-51B-Instruct-Q6_K.gguf	Q6_K	6	42.3 GB	very large, extremely low quality loss
Llama-3_1-Nemotron-51B-Instruct-Q8_0-00001-of-00002.gguf	Q8_0	8	29.8 GB	very large, extremely low quality loss - not recommended
Llama-3_1-Nemotron-51B-Instruct-Q8_0-00002-of-00002.gguf	Q8_0	8	24.9 GB	very large, extremely low quality loss - not recommended
Llama-3_1-Nemotron-51B-Instruct-f16-00001-of-00004.gguf	f16	16	29.8 GB
Llama-3_1-Nemotron-51B-Instruct-f16-00002-of-00004.gguf	f16	16	29.6 GB
Llama-3_1-Nemotron-51B-Instruct-f16-00003-of-00004.gguf	f16	16	29.6 GB
Llama-3_1-Nemotron-51B-Instruct-f16-00004-of-00004.gguf	f16	16	14.1 GB

Quantized with llama.cpp b4381