SmolVLM2-2.2B-Instruct-GGUF
Original Model
HuggingFaceTB/SmolVLM2-2.2B-Instruct
Run with LlamaEdge
-
LlamaEdge version: v0.21.0 and above
-
Prompt template
-
Prompt type:
smol-vision -
Prompt string
<|im_start|> User: {user_message_1}<image> Assistant: {assistant_message_1} User: {user_message_2}<image> Assistant:
-
-
Context size:
2048 -
Run as LlamaEdge service
wasmedge --dir .:. --nn-preload default:GGML:AUTO:SmolVLM2-2.2B-Instruct-Q5_K_M.gguf \ llama-api-server.wasm \ --prompt-template smol-vision \ --llava-mmproj SmolVLM2-2.2B-Instruct-mmproj-f16.gguf \ --model-name SmolVLM2-2.2B-Instruct \ --ctx-size 2048
Quantized GGUF Models
| Name | Quant method | Bits | Size | Use case |
|---|---|---|---|---|
| SmolVLM2-2.2B-Instruct-Q2_K.gguf | Q2_K | 2 | 708 MB | smallest, significant quality loss - not recommended for most purposes |
| SmolVLM2-2.2B-Instruct-Q3_K_L.gguf | Q3_K_L | 3 | 976 MB | small, substantial quality loss |
| SmolVLM2-2.2B-Instruct-Q3_K_M.gguf | Q3_K_M | 3 | 904 MB | very small, high quality loss |
| SmolVLM2-2.2B-Instruct-Q3_K_S.gguf | Q3_K_S | 3 | 820 MB | very small, high quality loss |
| SmolVLM2-2.2B-Instruct-Q4_0.gguf | Q4_0 | 4 | 1.05 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| SmolVLM2-2.2B-Instruct-Q4_K_M.gguf | Q4_K_M | 4 | 1.11 GB | medium, balanced quality - recommended |
| SmolVLM2-2.2B-Instruct-Q4_K_S.gguf | Q4_K_S | 4 | 1.06 GB | small, greater quality loss |
| SmolVLM2-2.2B-Instruct-Q5_0.gguf | Q5_0 | 5 | 1.26 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| SmolVLM2-2.2B-Instruct-Q5_K_M.gguf | Q5_K_M | 5 | 1.30 GB | large, very low quality loss - recommended |
| SmolVLM2-2.2B-Instruct-Q5_K_S.gguf | Q5_K_S | 5 | 1.26 GB | large, low quality loss - recommended |
| SmolVLM2-2.2B-Instruct-Q6_K.gguf | Q6_K | 6 | 1.49 GB | very large, extremely low quality loss |
| SmolVLM2-2.2B-Instruct-Q8_0.gguf | Q8_0 | 8 | 1.93 GB | very large, extremely low quality loss - not recommended |
| SmolVLM2-2.2B-Instruct-f16.gguf | f16 | 16 | 3.63 GB | |
| SmolVLM2-2.2B-Instruct-mmproj-f16.gguf | f16 | 16 | 872 MB |
Quantized with llama.cpp b5501