README.md

metadata

license: other
license_name: exaone
license_link: LICENSE
model_name: EXAONE-Deep-2.4B
base_model: LGAI-EXAONE/EXAONE-Deep-2.4B
quantized_by: Second State Inc.
language:
  - en
  - ko
tags:
  - lg-ai
  - exaone
  - gguf
pipeline_tag: text-generation
library_name: transformers

---

EXAONE-Deep-2.4B-GGUF

Original Model

LGAI-EXAONE/EXAONE-Deep-2.4B

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-2.4B-Q5_K_M.gguf \
      llama-api-server.wasm \
      --prompt-template exaone-deep-chat \
      --ctx-size 32000 \
      --model-name EXAONE-Deep-2.4B
  

• Run as LlamaEdge command app

  wasmedge --dir .:. --nn-preload default:GGML:AUTO:EXAONE-Deep-2.4B-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template exaone-chat \
    --ctx-size 32000
  

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
EXAONE-Deep-2.4B-Q2_K.gguf	Q2_K	2	1.01 GB	smallest, significant quality loss - not recommended for most purposes
EXAONE-Deep-2.4B-Q3_K_L.gguf	Q3_K_L	3	1.35 GB	small, substantial quality loss
EXAONE-Deep-2.4B-Q3_K_M.gguf	Q3_K_M	3	1.25 GB	very small, high quality loss
EXAONE-Deep-2.4B-Q3_K_S.gguf	Q3_K_S	3	1.14 GB	very small, high quality loss
EXAONE-Deep-2.4B-Q4_0.gguf	Q4_0	4	1.43 GB	legacy; small, very high quality loss - prefer using Q3_K_M
EXAONE-Deep-2.4B-Q4_K_M.gguf	Q4_K_M	4	1.50 GB	medium, balanced quality - recommended
EXAONE-Deep-2.4B-Q4_K_S.gguf	Q4_K_S	4	1.43 GB	small, greater quality loss
EXAONE-Deep-2.4B-Q5_0.gguf	Q5_0	5	1.69 GB	legacy; medium, balanced quality - prefer using Q4_K_M
EXAONE-Deep-2.4B-Q5_K_M.gguf	Q5_K_M	5	1.73 GB	large, very low quality loss - recommended
EXAONE-Deep-2.4B-Q5_K_S.gguf	Q5_K_S	5	1.69 GB	large, low quality loss - recommended
EXAONE-Deep-2.4B-Q6_K.gguf	Q6_K	6	1.98 GB	very large, extremely low quality loss
EXAONE-Deep-2.4B-Q8_0.gguf	Q8_0	8	2.56 GB	very large, extremely low quality loss - not recommended
EXAONE-Deep-2.4B-f16.gguf	f16	16	4.82 GB

Quantized with llama.cpp b4920.