Instructions to use cyttic/exp26-composed-stage1-frozen with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cyttic/exp26-composed-stage1-frozen with Transformers:

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

pipe = pipeline("image-text-to-text", model="cyttic/exp26-composed-stage1-frozen")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("cyttic/exp26-composed-stage1-frozen")
model = AutoModelForMultimodalLM.from_pretrained("cyttic/exp26-composed-stage1-frozen", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cyttic/exp26-composed-stage1-frozen with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cyttic/exp26-composed-stage1-frozen"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cyttic/exp26-composed-stage1-frozen",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cyttic/exp26-composed-stage1-frozen

SGLang

How to use cyttic/exp26-composed-stage1-frozen 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 "cyttic/exp26-composed-stage1-frozen" \
    --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": "cyttic/exp26-composed-stage1-frozen",
		"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 "cyttic/exp26-composed-stage1-frozen" \
        --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": "cyttic/exp26-composed-stage1-frozen",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use cyttic/exp26-composed-stage1-frozen with Docker Model Runner:
```
docker model run hf.co/cyttic/exp26-composed-stage1-frozen
```

exp26-composed-stage1-frozen

This model is a fine-tuned version of cyttic/exp2-frozen-benyehuda-cont on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 2.4017
Cer: 0.3913
Wer: 0.6601

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: 5e-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
lr_scheduler_warmup_steps: 0.05
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
4.0410	0.1533	2500	3.6620	0.5444	0.8565
3.4861	0.3065	5000	3.2243	0.4949	0.7929
3.2082	0.4598	7500	2.8709	0.4577	0.7558
2.8683	0.6130	10000	2.7565	0.4322	0.7055
2.6964	0.7663	12500	2.5223	0.4108	0.6962
2.6825	0.9195	15000	2.4180	0.3921	0.6624
2.6582	1.0	16313	2.4017	0.3913	0.6601

Framework versions

Transformers 5.12.1
Pytorch 2.11.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

Downloads last month: 7

Safetensors

Model size

0.3B params

Tensor type

F32

Model tree for cyttic/exp26-composed-stage1-frozen

Base model

cyttic/exp2-frozen-benyehuda-cont

Finetuned

(51)

this model