SmolLM-360M-Instruct

Original Model

Run with LlamaEdge

LlamaEdge version: v0.12.5 and above

Prompt template

Prompt type: chatml

Prompt string

<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Context size: 2048

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:SmolLM-360M-Instruct-Q5_K_M.gguf \
  llama-api-server.wasm \
  --prompt-template chatml \
  --ctx-size 2048 \
  --model-name SmolLM-360M-Instruct

Run as LlamaEdge command app

wasmedge --dir .:. --nn-preload default:GGML:AUTO:SmolLM-360M-Instruct-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template chatml \
  --ctx-size 2048

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
SmolLM-360M-Instruct-Q2_K.gguf	Q2_K	2	219 MB	smallest, significant quality loss - not recommended for most purposes
SmolLM-360M-Instruct-Q3_K_L.gguf	Q3_K_L	3	246 MB	small, substantial quality loss
SmolLM-360M-Instruct-Q3_K_M.gguf	Q3_K_M	3	235 MB	very small, high quality loss
SmolLM-360M-Instruct-Q3_K_S.gguf	Q3_K_S	3	219 MB	very small, high quality loss
SmolLM-360M-Instruct-Q4_0.gguf	Q4_0	4	229 MB	legacy; small, very high quality loss - prefer using Q3_K_M
SmolLM-360M-Instruct-Q4_K_M.gguf	Q4_K_M	4	271 MB	medium, balanced quality - recommended
SmolLM-360M-Instruct-Q4_K_S.gguf	Q4_K_S	4	260 MB	small, greater quality loss
SmolLM-360M-Instruct-Q5_0.gguf	Q5_0	5	268 MB	legacy; medium, balanced quality - prefer using Q4_K_M
SmolLM-360M-Instruct-Q5_K_M.gguf	Q5_K_M	5	290 MB	large, very low quality loss - recommended
SmolLM-360M-Instruct-Q5_K_S.gguf	Q5_K_S	5	283 MB	large, low quality loss - recommended
SmolLM-360M-Instruct-Q6_K.gguf	Q6_K	6	367 MB	very large, extremely low quality loss
SmolLM-360M-Instruct-Q8_0.gguf	Q8_0	8	386 MB	very large, extremely low quality loss - not recommended
SmolLM-360M-Instruct-f16.gguf	f16	16	726 MB

Quantized with llama.cpp b3445.