awq model upload

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+test_screenshots/amazon.png filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

added_tokens.json

@@ -0,0 +1,16 @@
+{
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

chat_template.json

@@ -0,0 +1,3 @@
+{
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}"
+}

config.json

@@ -0,0 +1,59 @@
+{
+  "_name_or_path": "/users/difei/mingyu/showui_local/showui-2b_safetensors",
+  "architectures": [
+    "Qwen2VLForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "image_token_id": 151655,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen2_vl",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "quantization_config": {
+    "bits": 4,
+    "group_size": 128,
+    "modules_to_not_convert": [
+      "visual"
+    ],
+    "quant_method": "awq",
+    "version": "gemm",
+    "zero_point": true
+  },
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": {
+    "mrope_section": [
+      16,
+      24,
+      24
+    ],
+    "rope_type": "default",
+    "type": "default"
+  },
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": true,
+  "tokenizer_model_max_length": 4096,
+  "torch_dtype": "float16",
+  "transformers_version": "4.47.0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "hidden_size": 1536,
+    "in_chans": 3,
+    "model_type": "qwen2_vl",
+    "spatial_patch_size": 14
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652,
+  "vision_token_id": 151654,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "_attn_implementation": "eager",
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.01,
+  "top_k": 1,
+  "top_p": 0.001,
+  "transformers_version": "4.47.0"
+}

infer.py

@@ -0,0 +1,162 @@
+import argparse
+import ast
+import torch
+from PIL import Image
+import time
+from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
+from PIL import Image, ImageDraw
+
+# Parse command-line arguments
+parser = argparse.ArgumentParser(description="Qwen-VL Inference")
+parser.add_argument("--image_path", type=str, required=False, default="./test_screenshots/amazon.png", help="Path to the input image")
+parser.add_argument("--query", type=str, required=False, default="Click on the 'Chairs'.", help="Text query or instruction")
+parser.add_argument("--model_dir", type=str, default="./showui-2b-awq-4bit", help="Path to the local ShowUI model directory")
+args = parser.parse_args()
+
+DEVICE = "cuda:0"
+
+# Constants for the processor (adjust as per your model requirements)
+MIN_PIXELS = 256 * 28 * 28
+# MAX_PIXELS = 1024 * 28 * 28
+# MAX_PIXELS = 1280 * 28 * 28
+MAX_PIXELS = 1344 * 28 * 28
+
+
+def draw_point_on_image(image_path, position, output_path="output_image.png", radius=2, color="red"):
+    image = Image.open(image_path).convert("RGB")
+    draw = ImageDraw.Draw(image)
+    width, height = image.size
+    x = int(position[0] * width)
+    y = int(position[1] * height)
+    draw.ellipse([(x - radius, y - radius), (x + radius, y + radius)], fill=color, outline=color)
+    image.save(output_path)
+    # print(f"Point drawn at ({x}, {y}) and saved to {output_path}")
+
+# Load model and processor
+model = Qwen2VLForConditionalGeneration.from_pretrained(
+    args.model_dir,
+    torch_dtype=torch.float16,
+    # torch_dtype=torch.bfloat16,
+    device_map="cpu",
+)
+
+print("Model dtype:", model.dtype)
+# print("Model weights dtype:", model.model.layers[0].self_attn.q_proj.weight)
+
+processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", min_pixels=MIN_PIXELS, max_pixels=MAX_PIXELS)
+
+# Print CUDA memory usage after model load
+torch.cuda.empty_cache()
+
+_NAV_SYSTEM = """You are an assistant trained to navigate the {_APP} screen. 
+    Given a task instruction, a screen observation, and an action history sequence, 
+    output the next action and wait for the next observation. 
+    Here is the action space:
+    {_ACTION_SPACE}
+    """
+    
+_ACTION_MAP = """
+        1. CLICK: Click on an element, value is not applicable and the position [x,y] is required. 
+        2. INPUT: Type a string into an element, value is a string to type and the position [x,y] is required. 
+        3. HOVER: Hover on an element, value is not applicable and the position [x,y] is required.
+        4. ENTER: Enter operation, value and position are not applicable.
+        5. SCROLL: Scroll the screen, value is the direction to scroll and the position is not applicable.
+        6. ESC: ESCAPE operation, value and position are not applicable.
+        7. PRESS: Long click on an element, value is not applicable and the position [x,y] is required. 
+        """
+        
+_SYSTEM = _NAV_SYSTEM.format(
+            _APP="web",
+            _ACTION_SPACE=_ACTION_MAP
+        )
+
+if args.query:
+    _QUERY = args.query
+else:
+    _QUERY = "Click on the 'Chairs'."
+
+# Construct the input message
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {"type": "text", "text": _SYSTEM},
+            {"type": "image", "image": args.image_path, "min_pixels": MIN_PIXELS, "max_pixels": MAX_PIXELS},
+            {"type": "text", "text": _QUERY}
+        ],
+    }
+]
+
+# Process the message through the processor
+text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+
+# The processor automatically handles image loading when `images` parameter is used
+image = Image.open(args.image_path).convert("RGB")
+inputs = processor(
+    text=[text],
+    images=[image],
+    padding=True,
+    return_tensors="pt"
+)
+
+# Move inputs and model to GPU
+model = model.to(DEVICE)
+inputs = inputs.to(DEVICE)
+
+# print("Model dtype after to(DEVICE):", model.model.layers[0].self_attn.q_proj.weight)
+print(f"Max CUDA memory after model load: {torch.cuda.max_memory_allocated(device=DEVICE)/1024**2:.2f} MB")
+
+# Reset memory stats before inference
+torch.cuda.reset_peak_memory_stats(device=DEVICE)
+
+N_RUNS = 10
+times = []
+
+model.eval()
+with torch.no_grad():
+
+    for i in range(N_RUNS):
+        start_time = time.time()
+        
+        generated_ids = model.generate(**inputs, max_new_tokens=128)
+
+        # Extract the portion of generated_ids corresponding to the new generation
+        generated_ids_trimmed = [out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)]
+        output_text = processor.batch_decode(
+            generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
+        )[0]
+
+        # Convert output string to Python object if it's a coordinate
+        try:
+            result = ast.literal_eval(output_text)
+        except:
+            result = output_text
+            
+        end_time = time.time()
+        times.append(end_time - start_time)
+        print(f"Run {i+1}/{N_RUNS} - Time: {end_time - start_time:.4f} s, Output: {result}")
+        
+        # output_str = "{'action': 'CLICK', 'value': None, 'position': [0.28, 0.29]}"
+        # parsed_output = ast.literal_eval(result)
+        if result['action'].upper() == 'CLICK':
+            x, y = result['position'][0], result['position'][1]
+            draw_point_on_image(args.image_path, [x, y], output_path="./output_image.png")
+        
+
+
+avg_time = sum(times) / len(times)
+print(f"Average per inference time: {avg_time:.4f} seconds")
+
+# Print CUDA memory usage after inference
+print(f"Max CUDA memory after inference: {torch.cuda.max_memory_allocated(device=DEVICE)/1024**2:.2f} MB")
+
+# Print image size information
+print(f"Input image size: {Image.open(args.image_path).size}")
+
+
+
+
+
+
+
+

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31600e0be35beae362bc1cb3333486bbff0ba645685d8e7c87ede080093b7b15
+size 2478389712

preprocessor_config.json

@@ -0,0 +1,29 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "max_pixels": 1053696,
+  "merge_size": 2,
+  "min_pixels": 200704,
+  "patch_size": 14,
+  "processor_class": "Qwen2VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "max_pixels": 12845056,
+    "min_pixels": 3136
+  },
+  "temporal_patch_size": 2
+}

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88a3a6fcb80132f76da8aa40cdc3fccd7e5d8468ef15421f5b0c2715e85217d2
+size 11420538

tokenizer_config.json

@@ -0,0 +1,147 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "max_pixels": 1053696,
+  "min_pixels": 200704,
+  "model_max_length": 32768,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "processor_class": "Qwen2VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}