Instructions to use trl-internal-testing/tiny-Qwen2_5_VLForConditionalGeneration with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trl-internal-testing/tiny-Qwen2_5_VLForConditionalGeneration 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-Qwen2_5_VLForConditionalGeneration")
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-Qwen2_5_VLForConditionalGeneration")
model = AutoModelForImageTextToText.from_pretrained("trl-internal-testing/tiny-Qwen2_5_VLForConditionalGeneration")
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-Qwen2_5_VLForConditionalGeneration 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-Qwen2_5_VLForConditionalGeneration"
# 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-Qwen2_5_VLForConditionalGeneration",
		"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-Qwen2_5_VLForConditionalGeneration

SGLang

How to use trl-internal-testing/tiny-Qwen2_5_VLForConditionalGeneration 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-Qwen2_5_VLForConditionalGeneration" \
    --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-Qwen2_5_VLForConditionalGeneration",
		"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-Qwen2_5_VLForConditionalGeneration" \
        --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-Qwen2_5_VLForConditionalGeneration",
		"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-Qwen2_5_VLForConditionalGeneration with Docker Model Runner:
```
docker model run hf.co/trl-internal-testing/tiny-Qwen2_5_VLForConditionalGeneration
```

Upload Qwen2_5_VLForConditionalGeneration

by qgallouedec HF Staff - opened 1 day ago

base: refs/heads/main

←

from: refs/pr/9

Discussion Files changed

+51

-19

Files changed (6) hide show

config.json +45 -12
generation_config.json +1 -1
model.safetensors +1 -1
preprocessor_config.json +0 -2
tokenizer.json +2 -2
video_preprocessor_config.json +2 -1

config.json CHANGED Viewed

@@ -7,22 +7,21 @@
   "dtype": "bfloat16",
   "eos_token_id": 151645,
   "hidden_act": "silu",
-  "hidden_size": 2048,
   "image_token_id": 151655,
   "initializer_range": 0.02,
   "intermediate_size": 11008,
   "max_position_embeddings": 128000,
   "max_window_layers": 70,
   "model_type": "qwen2_5_vl",
-  "num_attention_heads": 16,
-  "num_hidden_layers": 36,
   "num_key_value_heads": 2,
   "rms_norm_eps": 1e-06,
   "rope_scaling": {
     "mrope_section": [
-      16,
-      24,
-      24
     ],
     "rope_type": "default",
     "type": "default"
@@ -39,9 +38,44 @@
     "eos_token_id": 151645,
     "hidden_act": "silu",
     "hidden_size": 16,
     "initializer_range": 0.02,
     "intermediate_size": 11008,
     "layer_types": [
       "full_attention",
       "full_attention"
     ],
@@ -65,16 +99,18 @@
     "tie_word_embeddings": true,
     "use_cache": true,
     "use_sliding_window": false,
     "vision_token_id": 151654,
     "vocab_size": 151936
   },
-  "transformers_version": "4.57.5",
   "use_cache": true,
   "use_sliding_window": false,
   "video_token_id": 151656,
   "vision_config": {
     "depth": 2,
-    "embed_dim": 64,
     "fullatt_block_indexes": [
       7,
       15,
@@ -88,10 +124,7 @@
     "initializer_range": 0.02,
     "intermediate_size": 3420,
     "model_type": "qwen2_5_vl",
-    "num_attention_heads": 4,
-    "num_heads": 16,
-    "num_hidden_layers": 2,
-    "num_key_value_heads": 2,
     "out_hidden_size": 16,
     "patch_size": 14,
     "spatial_merge_size": 2,

   "dtype": "bfloat16",
   "eos_token_id": 151645,
   "hidden_act": "silu",
+  "hidden_size": 16,
   "image_token_id": 151655,
   "initializer_range": 0.02,
   "intermediate_size": 11008,
   "max_position_embeddings": 128000,
   "max_window_layers": 70,
   "model_type": "qwen2_5_vl",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 2,
   "num_key_value_heads": 2,
   "rms_norm_eps": 1e-06,
   "rope_scaling": {
     "mrope_section": [
+      1,
+      1
     ],
     "rope_type": "default",
     "type": "default"
     "eos_token_id": 151645,
     "hidden_act": "silu",
     "hidden_size": 16,
+    "image_token_id": null,
     "initializer_range": 0.02,
     "intermediate_size": 11008,
     "layer_types": [
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention",
       "full_attention",
       "full_attention"
     ],
     "tie_word_embeddings": true,
     "use_cache": true,
     "use_sliding_window": false,
+    "video_token_id": null,
+    "vision_end_token_id": 151653,
+    "vision_start_token_id": 151652,
     "vision_token_id": 151654,
     "vocab_size": 151936
   },
+  "transformers_version": "4.56.2",
   "use_cache": true,
   "use_sliding_window": false,
   "video_token_id": 151656,
   "vision_config": {
     "depth": 2,
     "fullatt_block_indexes": [
       7,
       15,
     "initializer_range": 0.02,
     "intermediate_size": 3420,
     "model_type": "qwen2_5_vl",
+    "num_heads": 4,
     "out_hidden_size": 16,
     "patch_size": 14,
     "spatial_merge_size": 2,

generation_config.json CHANGED Viewed

@@ -8,5 +8,5 @@
   "pad_token_id": 151643,
   "repetition_penalty": 1.05,
   "temperature": 1e-06,
-  "transformers_version": "4.57.5"
 }

   "pad_token_id": 151643,
   "repetition_penalty": 1.05,
   "temperature": 1e-06,
+  "transformers_version": "4.56.2"
 }

model.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff3a6ae1cca0754abbf5228d25e94a31a992b31d20a1a1f13d07c3eb9d33bd83
 size 7721168

 version https://git-lfs.github.com/spec/v1
+oid sha256:64bcc6a5ae04fcb0c61164070b95350f722e612ba11985da760121de2e324f63
 size 7721168

preprocessor_config.json CHANGED Viewed

@@ -7,7 +7,6 @@
   "do_center_crop": null,
   "do_convert_rgb": true,
   "do_normalize": true,
-  "do_pad": null,
   "do_rescale": true,
   "do_resize": true,
   "image_mean": [
@@ -25,7 +24,6 @@
   "max_pixels": 12845056,
   "merge_size": 2,
   "min_pixels": 3136,
-  "pad_size": null,
   "patch_size": 14,
   "processor_class": "Qwen2_5_VLProcessor",
   "resample": 3,

   "do_center_crop": null,
   "do_convert_rgb": true,
   "do_normalize": true,
   "do_rescale": true,
   "do_resize": true,
   "image_mean": [
   "max_pixels": 12845056,
   "merge_size": 2,
   "min_pixels": 3136,
   "patch_size": 14,
   "processor_class": "Qwen2_5_VLProcessor",
   "resample": 3,

tokenizer.json CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
-size 11421896

 version https://git-lfs.github.com/spec/v1
+oid sha256:ba0c439f7be467bf47d12a7e6f9adc6116201056fc60c67f431c679b7c16afc8
+size 11422064

video_preprocessor_config.json CHANGED Viewed

@@ -6,6 +6,7 @@
   "do_center_crop": null,
   "do_convert_rgb": true,
   "do_normalize": true,
   "do_rescale": true,
   "do_resize": true,
   "do_sample_frames": false,
@@ -27,7 +28,6 @@
   "min_frames": 4,
   "min_pixels": 3136,
   "num_frames": null,
-  "pad_size": null,
   "patch_size": 14,
   "processor_class": "Qwen2_5_VLProcessor",
   "resample": 3,
@@ -37,6 +37,7 @@
     "longest_edge": 12845056,
     "shortest_edge": 3136
   },
   "temporal_patch_size": 2,
   "video_metadata": null,
   "video_processor_type": "Qwen2VLVideoProcessor"

   "do_center_crop": null,
   "do_convert_rgb": true,
   "do_normalize": true,
+  "do_pad": null,
   "do_rescale": true,
   "do_resize": true,
   "do_sample_frames": false,
   "min_frames": 4,
   "min_pixels": 3136,
   "num_frames": null,
   "patch_size": 14,
   "processor_class": "Qwen2_5_VLProcessor",
   "resample": 3,
     "longest_edge": 12845056,
     "shortest_edge": 3136
   },
+  "size_divisor": null,
   "temporal_patch_size": 2,
   "video_metadata": null,
   "video_processor_type": "Qwen2VLVideoProcessor"