Qwen2.5-7B-Instruct-1M-GGUF

Original Model

Run with LlamaEdge

LlamaEdge version: v0.16.2

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: 1010000 (generation 8192 tokens)

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Qwen2.5-7B-Instruct-1M-Q5_K_M.gguf \
  llama-api-server.wasm \
  --model-name Qwen2.5-7B-Instruct-1M \
  --prompt-template chatml \
  --ctx-size 1010000

Run as LlamaEdge command app

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Qwen2.5-7B-Instruct-1M-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template chatml \
  --ctx-size 1010000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Qwen2.5-7B-Instruct-1M-Q2_K.gguf	Q2_K	2	3.02 GB	smallest, significant quality loss - not recommended for most purposes
Qwen2.5-7B-Instruct-1M-Q3_K_L.gguf	Q3_K_L	3	4.09 GB	small, substantial quality loss
Qwen2.5-7B-Instruct-1M-Q3_K_M.gguf	Q3_K_M	3	3.81 GB	very small, high quality loss
Qwen2.5-7B-Instruct-1M-Q3_K_S.gguf	Q3_K_S	3	3.49 GB	very small, high quality loss
Qwen2.5-7B-Instruct-1M-Q4_0.gguf	Q4_0	4	4.43 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Qwen2.5-7B-Instruct-1M-Q4_K_M.gguf	Q4_K_M	4	4.68 GB	medium, balanced quality - recommended
Qwen2.5-7B-Instruct-1M-Q4_K_S.gguf	Q4_K_S	4	4.46 GB	small, greater quality loss
Qwen2.5-7B-Instruct-1M-Q5_0.gguf	Q5_0	5	5.32 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Qwen2.5-7B-Instruct-1M-Q5_K_M.gguf	Q5_K_M	5	5.44 GB	large, very low quality loss - recommended
Qwen2.5-7B-Instruct-1M-Q5_K_S.gguf	Q5_K_S	5	5.32 GB	large, low quality loss - recommended
Qwen2.5-7B-Instruct-1M-Q6_K.gguf	Q6_K	6	6.25 GB	very large, extremely low quality loss
Qwen2.5-7B-Instruct-1M-Q8_0.gguf	Q8_0	8	8.21 GB	very large, extremely low quality loss - not recommended
Qwen2.5-7B-Instruct-1M-f16.gguf	f16	16	15.2 GB

Quantized with llama.cpp b4466