LLM EXPLORER 61,491 MODELS INDEXED

Qwen2 VL 7B Instruct GPTQ Int4 by Qwen

By Qwen · 2588 downloads
HF Repo ↗ Copy URL

Qwen2 VL 7B Instruct GPTQ Int4 is an open-source language model by Qwen. Features: 7b LLM, VRAM: 7GB, Context: 32K, License: apache-2.0, Quantized, Instruction-Based, LLM Explorer Score: 0.16.

  4-bit Base model:quantized:qwen/qwen... Base model:qwen/qwen2-vl-7b-in...   En   Gptq   Image-text-to-text   Instruct   Multimodal   Quantized   Qwen2 vl   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct-GPTQ-Int4 

Qwen2 VL 7B Instruct GPTQ Int4 Parameters and Internals

Model Type 
multimodal, text generation
Use Cases 
Areas:
Research, Commercial applications
Applications:
Visual understanding, Video-based question answering, Mobile device operation, Multilingual text understanding
Primary Use Cases:
Image and video processing, Complex task automation, Multilingual text recognition
Limitations:
Does not understand audio, Data timeliness limited to June 2023, Limited in recognizing individuals or IPs, Needs improved capability for complex instructions, Accuracy issues in object counting, Weak in 3D spatial reasoning
Supported Languages 
English (fluent), Chinese (fluent), European languages (basic), Japanese (basic), Korean (basic), Arabic (basic), Vietnamese (basic)
Training Details 
Data Sources:
Image datasets updated until June 2023
Model Architecture:
Naive Dynamic Resolution, Multimodal Rotary Position Embedding (M-ROPE)
Input Output 
Input Format:
Images, Videos, Text
Accepted Modalities:
Image, Video, Text
Output Format:
Text
Performance Tips:
Enable flash_attention_2 for better acceleration and memory saving, especially in multi-image and video scenarios.
LLM NameQwen2 VL 7B Instruct GPTQ Int4
Repository 🤗https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct-GPTQ-Int4 
Base Model(s)  Qwen2 VL 7B Instruct   Qwen/Qwen2-VL-7B-Instruct
Model Size7b
Required VRAM7 GB
Updated2024-09-08
MaintainerQwen
Model Typeqwen2_vl
Instruction-BasedYes
Model Files  4.0 GB: 1-of-2   3.0 GB: 2-of-2
Supported Languagesen
GPTQ QuantizationYes
Quantization Typegptq
Model ArchitectureQwen2VLForConditionalGeneration
Licenseapache-2.0
Context Length32768
Model Max Length32768
Transformers Version4.45.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size152064
Torch Data Typefloat16
Errorsreplace

Best Alternatives to Qwen2 VL 7B Instruct GPTQ Int4

Best Alternatives
Context / RAM
Downloads
Likes
Qwen2 VL 7B Instruct GPTQ Int832K / 10.2 GB126631
...L 7B Instruct Unsloth Bnb 4bit32K / 6.8 GB586013
OlmOCR 7B 0225 Preview 4bit32K / 5.6 GB221
...L 7B Instruct Abliterated 8bit32K / 8.9 GB291
Qwen2 VL 7B Instruct Bnb 4bit32K / 5.9 GB9366
Qwen2 VL 7B Instruct 4bit32K / 4.7 GB6482
Qwen2 VL 7B Instruct 8bit32K / 8.9 GB362
OlmOCR 7B 0225 Preview32K / 16.6 GB18655706
Qwen2 VL 7B Instruct32K / 16.7 GB3337
OlmOCR 7B 0225 Preview FP832K / 10 GB1279
Note: green Score (e.g. "73.2") means that the model is better than Qwen/Qwen2-VL-7B-Instruct-GPTQ-Int4.