File size: 1,084 Bytes
c987b62 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
---
language:
- multilingual
base_model:
- nanonets/Nanonets-OCR2-3B
tags:
- OCR
- image-to-text
- pdf2markdown
- VQA
pipeline_tag: image-text-to-text
library_name: transformers
---
Creation Code
```python
from transformers import AutoProcessor, Qwen2_5_VLForConditionalGeneration
from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import QuantizationModifier
from llmcompressor.utils import dispatch_for_generation
MODEL_ID = "nanonets/Nanonets-OCR2-3B"
# Load model.
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype="auto")
processor = AutoProcessor.from_pretrained(MODEL_ID)
# Configure the quantization algorithm and scheme.
# In this case, we:
# * quantize the weights to fp8 with per channel via ptq
# * quantize the activations to fp8 with dynamic per token
recipe = QuantizationModifier(
targets="Linear",
scheme="FP8_DYNAMIC",
ignore=["lm_head", "re:visual.*", "re:model.visual.*"],
)
# Apply quantization and save to disk in compressed-tensors format.
oneshot(model=model, recipe=recipe)
``` |