metadata

base_model: ByteDance-Seed/Seed-Coder-8B-Instruct
license: mit
model_creator: ByteDance-Seed
model_name: Seed-Coder-8B-Instruct
quantized_by: Second State Inc.

Seed-Coder-8B-Instruct-GGUF

Original Model

ByteDance-Seed/Seed-Coder-8B-Instruct

Run with LlamaEdge

LlamaEdge version: coming soon

Prompt template

Prompt type: seed-instruct

Prompt string

<[begin▁of▁sentence]>system
{system_message}

<[end▁of▁sentence]><[begin▁of▁sentence]>user
{user_message_1}<[end▁of▁sentence]><[begin▁of▁sentence]>assistant
{assistant_message_1}<[end▁of▁sentence]><[begin▁of▁sentence]>user
{user_message_2}<[end▁of▁sentence]><[begin▁of▁sentence]>assistant

Context size: 32000

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Seed-Coder-8B-Instruct-Q5_K_M.gguf \
  llama-api-server.wasm \
  --model-name Seed-Coder-8B-Instruct \
  --prompt-template seed-instruct \
  --ctx-size 32000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Seed-Coder-8B-Instruct-Q2_K.gguf	Q2_K	2	3.30 GB	smallest, significant quality loss - not recommended for most purposes
Seed-Coder-8B-Instruct-Q3_K_L.gguf	Q3_K_L	3	4.46 GB	small, substantial quality loss
Seed-Coder-8B-Instruct-Q3_K_M.gguf	Q3_K_M	3	4.15 GB	very small, high quality loss
Seed-Coder-8B-Instruct-Q3_K_S.gguf	Q3_K_S	3	3.80 GB	very small, high quality loss
Seed-Coder-8B-Instruct-Q4_0.gguf	Q4_0	4	4.81 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Seed-Coder-8B-Instruct-Q4_K_M.gguf	Q4_K_M	4	5.07 GB	medium, balanced quality - recommended
Seed-Coder-8B-Instruct-Q4_K_S.gguf	Q4_K_S	4	4.84 GB	small, greater quality loss
Seed-Coder-8B-Instruct-Q5_0.gguf	Q5_0	5	5.76 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Seed-Coder-8B-Instruct-Q5_K_M.gguf	Q5_K_M	5	5.90 GB	large, very low quality loss - recommended
Seed-Coder-8B-Instruct-Q5_K_S.gguf	Q5_K_S	5	5.76 GB	large, low quality loss - recommended
Seed-Coder-8B-Instruct-Q6_K.gguf	Q6_K	6	6.78 GB	very large, extremely low quality loss
Seed-Coder-8B-Instruct-Q8_0.gguf	Q8_0	8	8.77 GB	very large, extremely low quality loss - not recommended
Seed-Coder-8B-Instruct-f16.gguf	f16	16	16.6 GB

Quantized with llama.cpp b5341