Instructions to use SCreates/Modi-TrOCR-Final with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SCreates/Modi-TrOCR-Final with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="SCreates/Modi-TrOCR-Final")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("SCreates/Modi-TrOCR-Final")
model = AutoModelForMultimodalLM.from_pretrained("SCreates/Modi-TrOCR-Final", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SCreates/Modi-TrOCR-Final with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SCreates/Modi-TrOCR-Final"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SCreates/Modi-TrOCR-Final",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/SCreates/Modi-TrOCR-Final

SGLang

How to use SCreates/Modi-TrOCR-Final with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "SCreates/Modi-TrOCR-Final" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SCreates/Modi-TrOCR-Final",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "SCreates/Modi-TrOCR-Final" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SCreates/Modi-TrOCR-Final",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use SCreates/Modi-TrOCR-Final with Docker Model Runner:
```
docker model run hf.co/SCreates/Modi-TrOCR-Final
```

Modi-TrOCR-Final

This model is a fine-tuned version of SCreates/Modi-TrOCR-Synthetic on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.3125
Cer: 0.8972

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 15
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Cer	Validation Loss
3.7382	0.8696	100	0.8934	1.8319
3.5762	1.7391	200	0.9024	1.7493
3.4447	2.6087	300	0.8939	1.6899
3.3490	3.4783	400	0.8971	1.6324
3.2533	4.3478	500	1.6137	0.9022
3.1613	5.2174	600	1.5621	0.9080
3.0370	6.0870	700	1.5066	0.8984
2.9625	6.9565	800	1.4612	0.9013
2.8664	7.8261	900	1.4233	0.9017
2.8030	8.6957	1000	1.3973	0.9024
2.7140	9.5652	1100	1.3715	0.8967
2.6395	10.4348	1200	1.3604	0.8968
2.5984	11.3043	1300	1.3392	0.8972
2.5495	12.1739	1400	1.3325	0.8982
2.5302	13.0435	1500	1.3240	0.8974
2.5043	13.9130	1600	1.3132	0.8972
2.4682	14.7826	1700	1.3125	0.8972

Framework versions

Transformers 5.0.0
Pytorch 2.10.0+cu128
Datasets 4.8.3
Tokenizers 0.22.2

Downloads last month: 16

Safetensors

Model size

0.3B params

Tensor type

F32

Model tree for SCreates/Modi-TrOCR-Final

Base model

microsoft/trocr-base-handwritten

Finetuned

SCreates/Modi-TrOCR-Synthetic

Finetuned

(1)

this model