File size: 3,891 Bytes
23f48d8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1a7150e
 
be75a6c
23f48d8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
be75a6c
 
 
 
 
 
 
 
 
 
 
23f48d8
 
 
 
 
 
 
1a7150e
23f48d8
 
 
 
8852f54
23f48d8
 
1a7150e
 
be75a6c
1a7150e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
import torch


from transformers import AutoProcessor, AutoModelForVision2Seq, AwqConfig

import numpy as np
import pyttsx3


START_TO_COUCH = np.array([[0.5, 0], [0.5, 0.5]]).ravel()
COUCH_TO_KITCHEN = np.array([[0.5, -0.5], [1.0, -1.0]]).ravel()
KITCHEN_TO_START = np.array([[0.5, -0.5], [0, 0]]).ravel()

engine = pyttsx3.init("espeak")
voices = engine.getProperty("voices")
engine.setProperty("voice", voices[3].id)


def speak(text):
    print(f"said {text}", flush=True)
    engine.say(text)
    engine.runAndWait()


speak("hello")

MODE = "fused_quantized"
DEVICE = "cuda"
PROCESSOR = AutoProcessor.from_pretrained("HuggingFaceM4/idefics2-tfrm-compatible")
BAD_WORDS_IDS = PROCESSOR.tokenizer(
    ["<image>", "<fake_token_around_image>"], add_special_tokens=False
).input_ids
EOS_WORDS_IDS = PROCESSOR.tokenizer(
    "<end_of_utterance>", add_special_tokens=False
).input_ids + [PROCESSOR.tokenizer.eos_token_id]

# Load model
if MODE == "regular":
    model = AutoModelForVision2Seq.from_pretrained(
        "HuggingFaceM4/idefics2-tfrm-compatible",
        torch_dtype=torch.float16,
        trust_remote_code=True,
        _attn_implementation="flash_attention_2",
        revision="3dc93be345d64fb6b1c550a233fe87ddb36f183d",
    ).to(DEVICE)
elif MODE == "quantized":
    quant_path = "HuggingFaceM4/idefics2-tfrm-compatible-AWQ"
    model = AutoModelForVision2Seq.from_pretrained(
        quant_path, trust_remote_code=True
    ).to(DEVICE)
elif MODE == "fused_quantized":
    quant_path = "HuggingFaceM4/idefics2-tfrm-compatible-AWQ"
    quantization_config = AwqConfig(
        bits=4,
        fuse_max_seq_len=4096,
        modules_to_fuse={
            "attention": ["q_proj", "k_proj", "v_proj", "o_proj"],
            "mlp": ["gate_proj", "up_proj", "down_proj"],
            "layernorm": ["input_layernorm", "post_attention_layernorm", "norm"],
            "use_alibi": False,
            "num_attention_heads": 32,
            "num_key_value_heads": 8,
            "hidden_size": 4096,
        },
    )
    model = AutoModelForVision2Seq.from_pretrained(
        quant_path, quantization_config=quantization_config, trust_remote_code=True
    ).to(DEVICE)
else:
    raise ValueError("Unknown mode")


def reset_awq_cache(model):
    """
    Simple method to reset the AWQ fused modules cache
    """
    from awq.modules.fused.attn import QuantAttentionFused

    for name, module in model.named_modules():
        if isinstance(module, QuantAttentionFused):
            module.start_pos = 0


def ask_vlm(image, instruction):
    prompts = [
        "User:",
        image,
        f"{instruction}.<end_of_utterance>\n",
        "Assistant:",
    ]
    speak(instruction)
    inputs = PROCESSOR(prompts)
    inputs = {k: torch.tensor(v).to(DEVICE) for k, v in inputs.items()}

    generated_ids = model.generate(
        **inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=50
    )
    generated_texts = PROCESSOR.batch_decode(generated_ids, skip_special_tokens=True)

    text = generated_texts[0].split("\nAssistant: ")[1]
    reset_awq_cache(model)
    speak(text)
    return text


# import requests
# import torch
# from PIL import Image
# from io import BytesIO


# def download_image(url):
# try:
# # Send a GET request to the URL to download the image
# response = requests.get(url)
# # Check if the request was successful (status code 200)
# if response.status_code == 200:
# # Open the image using PIL
# image = Image.open(BytesIO(response.content))
# # Return the PIL image object
# return image
# else:
# print(f"Failed to download image. Status code: {response.status_code}")
# return None
# except Exception as e:
# print(f"An error occurred: {e}")
# return None


# # Create inputs
# image1 = download_image(
# "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
# )

# print(ask_vlm(image1, "What is this?"))