metadata

license: other
license_name: exaone
license_link: LICENSE
model_name: EXAONE-Deep-32B
base_model: LGAI-EXAONE/EXAONE-Deep-32B
quantized_by: Second State Inc.
language:
  - en
  - ko
tags:
  - lg-ai
  - exaone
  - exaone-deep
pipeline_tag: text-generation
library_name: transformers

EXAONE-Deep-32B-GGUF

Original Model

LGAI-EXAONE/EXAONE-Deep-32B

Run with LlamaEdge

LlamaEdge version: v0.16.12 and above

Prompt template

Prompt type: exaone-deep-chat

Prompt string

[|system|]{system_message}[|endofturn|]
[|user|]{user_message_1}
[|assistant|]{assistant_message_1}[|endofturn|]
[|user|]{user_message_2}
[|assistant|]<thought>

Context size: 32000

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:EXAONE-Deep-32B-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template exaone-deep-chat \
    --ctx-size 32000 \
    --model-name EXAONE-Deep-32B

Run as LlamaEdge command app

wasmedge --dir .:. --nn-preload default:GGML:AUTO:EXAONE-Deep-32B-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template exaone-deep-chat \
  --ctx-size 32000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
EXAONE-Deep-32B-Q2_K.gguf	Q2_K	2	11.9 GB	smallest, significant quality loss - not recommended for most purposes
EXAONE-Deep-32B-Q3_K_L.gguf	Q3_K_L	3	16.8 GB	small, substantial quality loss
EXAONE-Deep-32B-Q3_K_M.gguf	Q3_K_M	3	15.5 GB	very small, high quality loss
EXAONE-Deep-32B-Q3_K_S.gguf	Q3_K_S	3	14.0 GB	very small, high quality loss
EXAONE-Deep-32B-Q4_0.gguf	Q4_0	4	18.1 GB	legacy; small, very high quality loss - prefer using Q3_K_M
EXAONE-Deep-32B-Q4_K_M.gguf	Q4_K_M	4	19.3 GB	medium, balanced quality - recommended
EXAONE-Deep-32B-Q4_K_S.gguf	Q4_K_S	4	18.3 GB	small, greater quality loss
EXAONE-Deep-32B-Q5_0.gguf	Q5_0	5	22.1 GB	legacy; medium, balanced quality - prefer using Q4_K_M
EXAONE-Deep-32B-Q5_K_M.gguf	Q5_K_M	5	22.7 GB	large, very low quality loss - recommended
EXAONE-Deep-32B-Q5_K_S.gguf	Q5_K_S	5	22.1 GB	large, low quality loss - recommended
EXAONE-Deep-32B-Q6_K.gguf	Q6_K	6	26.3 GB	very large, extremely low quality loss
EXAONE-Deep-32B-Q8_0.gguf	Q8_0	8	34.0 GB	very large, extremely low quality loss - not recommended
EXAONE-Deep-32B-f16-00001-of-00003.gguf	f16	16	29.8 GB
EXAONE-Deep-32B-f16-00002-of-00003.gguf	f16	16	30.0 GB
EXAONE-Deep-32B-f16-00003-of-00003.gguf	f16	16	4.23 GB

Quantized with llama.cpp b4914.