eiby777/olmocr-2-7b-1025-fp8 🖼️🔢 → 📝

▶️ 79 runs 📅 Oct 2025 ⚙️ Cog 0.16.8 🔗 GitHub ⚖️ License

image-to-text ocr

About

Quantized to FP8 Version of olmOCR-2-7B-1025, using llmcompressor.

Example Output

Output

primary_language: en
is_rotation_valid: True
rotation_correction: 0
is_table: False
is_diagram: False

The following represents analytical data compiled from smoke analyses of segments 1 and 2 and also includes the physical filtration conducted on the filter rods.

Smoke Analysis

	No Filter	With Filter (Percent)
a. Tobacco Rod Burned, mm.	51	51
b. Puffs / Cig.	7.2	7.8	8.3
c. TPM (Wet), mg/cig.	24.0	16.4	31.7
d. Nicotine, mg/cig.	1.06	0.85	19.8
e. FTC "Tar", mg/cig.	20.3	14.2	30.1

Performance Metrics

10.15s Prediction Time

110.63s Total Time

All Input Parameters

{
  "page": 1,
  "image": "https://replicate.delivery/pbxt/Nya8v7gzoB2ZE1ErD3cz7MBG7fS9s7IWzGeVoWEneqhFZAl4/test.png",
  "target_dim": 1288
}

Input Parameters

pdf Type: string: Upload a PDF file for OCR processing
page Type: integerDefault: 1: PDF page number to process (1-indexed)
image Type: string: Upload an image (PNG, JPG, JPEG) for OCR processing
target_dim Type: integerDefault: 1288: Target longest dimension for processing

Output Schema

Output

Type: string

Example Execution Logs

🖼️ Processing image: tmpyjxdha7ztest.png
✅ OCR processing completed!

Version Details

Version ID: 2c72f940c02b446d8795f6ba9f5804f8a75c2e401d51b28d1cb3d8826aca15bc
Version Created: October 30, 2025

Run on Replicate →