Qwen2.5-3B-Instruct-GGUF
Original Model
Run with LlamaEdge
-
LlamaEdge version: v0.14.3 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:
32000 -
Run as LlamaEdge service
wasmedge --dir .:. --nn-preload default:GGML:AUTO:Qwen2.5-3B-Instruct-Q5_K_M.gguf \ llama-api-server.wasm \ --model-name Qwen2.5-3B-Instruct \ --prompt-template chatml \ --ctx-size 32000 -
Run as LlamaEdge command app
wasmedge --dir .:. --nn-preload default:GGML:AUTO:Qwen2.5-3B-Instruct-Q5_K_M.gguf \ llama-chat.wasm \ --prompt-template chatml \ --ctx-size 32000
Quantized GGUF Models
| Name | Quant method | Bits | Size | Use case |
|---|---|---|---|---|
| Qwen2.5-3B-Instruct-Q2_K.gguf | Q2_K | 2 | 676 MB | smallest, significant quality loss - not recommended for most purposes |
| Qwen2.5-3B-Instruct-Q3_K_L.gguf | Q3_K_L | 3 | 880 MB | small, substantial quality loss |
| Qwen2.5-3B-Instruct-Q3_K_M.gguf | Q3_K_M | 3 | 824 MB | very small, high quality loss |
| Qwen2.5-3B-Instruct-Q3_K_S.gguf | Q3_K_S | 3 | 761 MB | very small, high quality loss |
| Qwen2.5-3B-Instruct-Q4_0.gguf | Q4_0 | 4 | 935 MB | legacy; small, very high quality loss - prefer using Q3_K_M |
| Qwen2.5-3B-Instruct-Q4_K_M.gguf | Q4_K_M | 4 | 986 MB | medium, balanced quality - recommended |
| Qwen2.5-3B-Instruct-Q4_K_S.gguf | Q4_K_S | 4 | 940 MB | small, greater quality loss |
| Qwen2.5-3B-Instruct-Q5_0.gguf | Q5_0 | 5 | 1.1 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| Qwen2.5-3B-Instruct-Q5_K_M.gguf | Q5_K_M | 5 | 1.13 GB | large, very low quality loss - recommended |
| Qwen2.5-3B-Instruct-Q5_K_S.gguf | Q5_K_S | 5 | 1.1 GB | large, low quality loss - recommended |
| Qwen2.5-3B-Instruct-Q6_K.gguf | Q6_K | 6 | 1.27 GB | very large, extremely low quality loss |
| Qwen2.5-3B-Instruct-Q8_0.gguf | Q8_0 | 8 | 1.65 GB | very large, extremely low quality loss - not recommended |
| Qwen2.5-3B-Instruct-f16.gguf | f16 | 16 | 3.09 GB |
Quantized with llama.cpp b4466