Qwen2.5-VL-7B-Instruct

Original model repository: Qwen/Qwen2.5-VL-7B-Instruct

Model Introduction

Qwen2.5-VL-7B-Instruct is an instruction-tuned Vision-Language Model (VLM) for understanding images, videos, and text. It combines a vision transformer with the Qwen2.5 language model and supports visual question answering, OCR, document and chart analysis, visual localization, image description, video understanding, and visual-agent tasks.

Deployment Metrics

Model Parameters

Metric	Value
Total model parameters	8.292B
Vision model (ViT) parameters	676.6M
Language model (LM) parameters	7.616B

Parameter counts are calculated from the tensors stored in the upstream checkpoint.

Performance Metrics

Chips	Data Type	ViT Image Size	Sequence Length (tokens)	Maximum Context Length (tokens)	BPU Cores (ViT / Prefill / Decode)	ViT Latency (ms)	TTFT (ms)	Prefill TPS (token/s)	Decode TPS (token/s)	BPU Memory (GB)	CPU Memory (GB)
Matrix6P	W8A8	448 × 448	512	1024	4 / 4 / 4	43.234	200.656	3,336.058	21.190	8.3	2.2

Note: TTFT includes preprocessing and ViT latency. Memory values represent the peak memory usage measured during the specified performance test.