Instructions to use trl-internal-testing/tiny-Gemma3ForConditionalGeneration with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trl-internal-testing/tiny-Gemma3ForConditionalGeneration 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-Gemma3ForConditionalGeneration")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("trl-internal-testing/tiny-Gemma3ForConditionalGeneration")
model = AutoModelForImageTextToText.from_pretrained("trl-internal-testing/tiny-Gemma3ForConditionalGeneration")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use trl-internal-testing/tiny-Gemma3ForConditionalGeneration 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-Gemma3ForConditionalGeneration"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trl-internal-testing/tiny-Gemma3ForConditionalGeneration",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/trl-internal-testing/tiny-Gemma3ForConditionalGeneration

SGLang

How to use trl-internal-testing/tiny-Gemma3ForConditionalGeneration 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-Gemma3ForConditionalGeneration" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trl-internal-testing/tiny-Gemma3ForConditionalGeneration",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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-Gemma3ForConditionalGeneration" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trl-internal-testing/tiny-Gemma3ForConditionalGeneration",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use trl-internal-testing/tiny-Gemma3ForConditionalGeneration with Docker Model Runner:
```
docker model run hf.co/trl-internal-testing/tiny-Gemma3ForConditionalGeneration
```

qgallouedec HF Staff commited on Aug 13, 2025

Commit

ca18b3a

verified ·

1 Parent(s): 505d3a7

Upload Gemma3ForConditionalGeneration

Browse files

Files changed (3) hide show

config.json +50 -10
generation_config.json +4 -1
model.safetensors +2 -2

config.json CHANGED Viewed

@@ -4,6 +4,10 @@
   ],
   "boi_token_index": 255999,
   "eoi_token_index": 256000,
   "image_token_index": 262144,
   "initializer_range": 0.02,
   "mm_tokens_per_image": 256,
@@ -18,24 +22,59 @@
     "hidden_activation": "gelu_pytorch_tanh",
     "hidden_size": 16,
     "initializer_range": 0.02,
-    "intermediate_size": 32,
     "layer_types": [
       "sliding_attention",
       "sliding_attention"
     ],
     "max_position_embeddings": 131072,
     "model_type": "gemma3_text",
-    "num_attention_heads": 4,
     "num_hidden_layers": 2,
-    "num_key_value_heads": 2,
     "query_pre_attn_scalar": 256,
     "rms_norm_eps": 1e-06,
     "rope_local_base_freq": 10000.0,
-    "rope_scaling": null,
     "rope_theta": 1000000.0,
-    "sliding_window": 4096,
     "use_cache": true,
-    "vocab_size": 268559
   },
   "torch_dtype": "float32",
   "transformers_version": "4.56.0.dev0",
@@ -43,13 +82,14 @@
     "attention_dropout": 0.0,
     "hidden_act": "gelu_pytorch_tanh",
     "hidden_size": 16,
-    "image_size": 336,
-    "intermediate_size": 32,
     "layer_norm_eps": 1e-06,
     "model_type": "siglip_vision_model",
-    "num_attention_heads": 4,
     "num_channels": 3,
     "num_hidden_layers": 2,
-    "patch_size": 20
   }
 }

   ],
   "boi_token_index": 255999,
   "eoi_token_index": 256000,
+  "eos_token_id": [
+    1,
+    106
+  ],
   "image_token_index": 262144,
   "initializer_range": 0.02,
   "mm_tokens_per_image": 256,
     "hidden_activation": "gelu_pytorch_tanh",
     "hidden_size": 16,
     "initializer_range": 0.02,
+    "intermediate_size": 10240,
     "layer_types": [
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
       "sliding_attention",
       "sliding_attention"
     ],
     "max_position_embeddings": 131072,
     "model_type": "gemma3_text",
+    "num_attention_heads": 8,
     "num_hidden_layers": 2,
+    "num_key_value_heads": 4,
     "query_pre_attn_scalar": 256,
     "rms_norm_eps": 1e-06,
     "rope_local_base_freq": 10000.0,
+    "rope_scaling": {
+      "factor": 8.0,
+      "rope_type": "linear"
+    },
     "rope_theta": 1000000.0,
+    "sliding_window": 1024,
     "use_cache": true,
+    "vocab_size": 262208
   },
   "torch_dtype": "float32",
   "transformers_version": "4.56.0.dev0",
     "attention_dropout": 0.0,
     "hidden_act": "gelu_pytorch_tanh",
     "hidden_size": 16,
+    "image_size": 896,
+    "intermediate_size": 4304,
     "layer_norm_eps": 1e-06,
     "model_type": "siglip_vision_model",
+    "num_attention_heads": 16,
     "num_channels": 3,
     "num_hidden_layers": 2,
+    "patch_size": 14,
+    "vision_use_head": false
   }
 }

generation_config.json CHANGED Viewed

@@ -1,7 +1,10 @@
 {
   "_from_model_config": true,
   "bos_token_id": 2,
-  "eos_token_id": 1,
   "pad_token_id": 0,
   "transformers_version": "4.56.0.dev0"
 }

 {
   "_from_model_config": true,
   "bos_token_id": 2,
+  "eos_token_id": [
+    1,
+    106
+  ],
   "pad_token_id": 0,
   "transformers_version": "4.56.0.dev0"
 }

model.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:70e7e358822675eabb72cdeca6dc61b23d44706cdd77c86ff221249cae28c17c
-size 17728776

 version https://git-lfs.github.com/spec/v1
+oid sha256:db80e8136185f45ba7732fd1f88b0fc7446ff8169fafdded5cdf9f57a7303b70
+size 22959728