Instructions to use trl-internal-testing/tiny-PaliGemmaForConditionalGeneration with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trl-internal-testing/tiny-PaliGemmaForConditionalGeneration with Transformers:

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

pipe = pipeline("image-text-to-text", model="trl-internal-testing/tiny-PaliGemmaForConditionalGeneration")

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("trl-internal-testing/tiny-PaliGemmaForConditionalGeneration")
model = AutoModelForImageTextToText.from_pretrained("trl-internal-testing/tiny-PaliGemmaForConditionalGeneration")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use trl-internal-testing/tiny-PaliGemmaForConditionalGeneration with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/trl-internal-testing/tiny-PaliGemmaForConditionalGeneration

SGLang

How to use trl-internal-testing/tiny-PaliGemmaForConditionalGeneration 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 "trl-internal-testing/tiny-PaliGemmaForConditionalGeneration" \
    --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": "trl-internal-testing/tiny-PaliGemmaForConditionalGeneration",
		"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 "trl-internal-testing/tiny-PaliGemmaForConditionalGeneration" \
        --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": "trl-internal-testing/tiny-PaliGemmaForConditionalGeneration",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use trl-internal-testing/tiny-PaliGemmaForConditionalGeneration with Docker Model Runner:
```
docker model run hf.co/trl-internal-testing/tiny-PaliGemmaForConditionalGeneration
```

tiny-PaliGemmaForConditionalGeneration / config.json

qgallouedec HF Staff

Upload PaliGemmaForConditionalGeneration

2c7bec0 verified 5 months ago

raw

history blame contribute delete

1.44 kB

	{
	"architectures": [
	"PaliGemmaForConditionalGeneration"
	],
	"bos_token_id": 2,
	"dtype": "float32",
	"eos_token_id": 1,
	"hidden_size": 2048,
	"ignore_index": -100,
	"image_token_index": 257152,
	"model_type": "paligemma",
	"pad_token_id": 0,
	"projection_dim": 2048,
	"text_config": {
	"attention_bias": false,
	"attention_dropout": 0.0,
	"dtype": "float32",
	"head_dim": 256,
	"hidden_act": "gelu_pytorch_tanh",
	"hidden_activation": null,
	"hidden_size": 16,
	"initializer_range": 0.02,
	"intermediate_size": 16384,
	"layer_types": null,
	"max_position_embeddings": 8192,
	"model_type": "gemma",
	"num_attention_heads": 4,
	"num_hidden_layers": 2,
	"num_image_tokens": 256,
	"num_key_value_heads": 2,
	"rms_norm_eps": 1e-06,
	"rope_theta": 10000.0,
	"use_cache": true,
	"vocab_size": 257216
	},
	"transformers_version": "4.57.3",
	"vision_config": {
	"attention_dropout": 0.0,
	"embed_dim": 64,
	"hidden_act": "gelu_pytorch_tanh",
	"hidden_size": 16,
	"image_size": 224,
	"intermediate_size": 4304,
	"layer_norm_eps": 1e-06,
	"model_type": "siglip_vision_model",
	"num_attention_heads": 4,
	"num_channels": 3,
	"num_hidden_layers": 2,
	"num_image_tokens": 256,
	"num_key_value_heads": 2,
	"patch_size": 14,
	"projection_dim": 2048,
	"projector_hidden_act": "gelu_fast",
	"vision_use_head": false
	}
	}