SamMikaelson
/

deepseek-ocr-qvlm-4bit

deepseek-ocr-qvlm-4bit / quantization_config.json

Upload QVLM 4-bit quantized DeepSeek-OCR model (SafeTensors format) - 2.89x compression

cad72e6 verified 10 months ago

127 Bytes

{
  "quantization_bits": 4,
  "group_size": 128,
  "vision_quantize": true,
  "language_quantize": true,
  "symmetric": false
}