File size: 8,889 Bytes
11e7313
 
 
 
 
 
 
 
 
 
 
 
e608f06
11e7313
 
 
 
 
 
e608f06
 
 
 
6a1b0dd
 
 
 
 
 
 
11e7313
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e608f06
11e7313
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e608f06
11e7313
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
import os,random,math
import torch
import torch.nn as nn
from torchvision import transforms
from tqdm import tqdm
import open_clip
from datasets import load_dataset
from PIL import Image
from src.preprocessing import Preprocessing
from torch.utils.data import DataLoader,Dataset
import warnings
import base64
from huggingface_hub import hf_hub_download
from io import BytesIO
warnings.filterwarnings("ignore", message=".*QuickGELU mismatch.*")

device='cuda' if torch.cuda.is_available() else 'cpu'
torch.cuda.empty_cache()
model, _, preprocess =open_clip.create_model_and_transforms('ViT-B-32',pretrained='openai',device=device )
HF_TOKEN=os.getenv("HF_TOKEN")

MODEL_ID = "PrashantGoyal/findr-clip-ft"

model_path = hf_hub_download(
    repo_id=MODEL_ID,
    force_download=True,
    filename="clip/best.pt",
    token=os.getenv("HF_TOKEN")
)
# model_path = "model/clip/best.pt"
tokenizer=open_clip.get_tokenizer('ViT-B-32')

def seed_everything(seed=42):
    random.seed(seed); torch.manual_seed(seed); torch.cuda.manual_seed_all(seed)

class clip_dataset(torch.utils.data.Dataset):

    def __init__(self,split='val',processor=None,tokenizer=None):
        preprocessor=Preprocessing()
        self.ds=preprocessor.load_dataset(split=split)
        self.tokenizer=tokenizer
        self.processor=processor

    def __len__(self):
        return len(self.ds)    
    
    def __getitem__(self,index):
        data=self.ds[index]
        img:Image.Image=data['image'].convert('RGB')
        text=random.choice(data['answer']).strip()
        image=self.processor(img) if self.processor else img
        token_text=self.tokenizer([text])[0]
        return image,token_text
    
def clip_loss(image_features, text_features, temperature):
    image_features = image_features / image_features.norm(dim=-1, keepdim=True)
    text_features  = text_features  / text_features.norm(dim=-1, keepdim=True)
    logits_per_image = (image_features @ text_features.t()) * torch.exp(temperature)
    logits_per_text  = logits_per_image.t()
    targets = torch.arange(image_features.size(0), device=image_features.device)
    loss_i = nn.CrossEntropyLoss()(logits_per_image, targets)
    loss_t = nn.CrossEntropyLoss()(logits_per_text, targets)
    return (loss_i + loss_t) / 2    

def collate(batch):
        imgs, toks = zip(*batch)
        imgs = torch.stack(imgs, 0)
        toks = torch.stack(toks, 0)
        return imgs, toks            

def train(arch='ViT-B-32',pretrained='openai',batchSize=2,epochs=5,lr=5e-5,warmup_steps=200,grad_accum=1,output_dir='model/clip'):
    seed_everything(42)
    torch.cuda.empty_cache()
    os.makedirs(output_dir,exist_ok=True)
    
    tokenizer=open_clip.get_tokenizer(arch)

    train_ds=clip_dataset(split='val',processor=preprocess,tokenizer=tokenizer)
    val_ds=clip_dataset(split='test',processor=preprocess,tokenizer=tokenizer)

    train_dl = DataLoader(train_ds, batch_size=batchSize, shuffle=True, num_workers=4, collate_fn=collate, pin_memory=True)
    val_dl   = DataLoader(val_ds,   batch_size=batchSize, shuffle=False, num_workers=4, collate_fn=collate, pin_memory=True)

    total_steps = epochs * math.ceil(len(train_dl) / grad_accum)
    def lr_lambda(step):
        if step < warmup_steps:
            return (step + 1) / max(1, warmup_steps)
        progress = (step - warmup_steps) / max(1, total_steps - warmup_steps)
        return 0.5 * (1 + math.cos(math.pi * progress))
    

    optimizer = torch.optim.AdamW(model.parameters(), lr=lr, weight_decay=0.01)
    scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer, lr_lambda)

    scaler = torch.cuda.amp.GradScaler(enabled=(device.startswith("cuda")))
    best_val = float("inf")

    for epoch in range(1,epochs+1):
        model.train()
        running = 0.0
        step = 0
        pbar = tqdm(train_dl, desc=f"Epoch {epoch}/{epochs}")
        optimizer.zero_grad(set_to_none=True)
        for images, tokens in pbar:
            images = images.to(device, non_blocking=True)
            tokens = tokens.to(device, non_blocking=True)

            with torch.cuda.amp.autocast(enabled=(device.startswith("cuda"))):
                image_features = model.encode_image(images)
                text_features  = model.encode_text(tokens)
                loss = clip_loss(image_features, text_features, model.logit_scale)

            scaler.scale(loss / grad_accum).backward()
            step += 1
            running += loss.item()
            if step % grad_accum == 0:
                scaler.step(optimizer); scaler.update()
                optimizer.zero_grad(set_to_none=True)
                scheduler.step()

            pbar.set_postfix(loss=running / step, lr=optimizer.param_groups[0]["lr"])


    model.eval()
    with torch.no_grad():
            val_losses = []
            for images, tokens in tqdm(val_dl, leave=False, desc="Val"):
                images = images.to(device); tokens = tokens.to(device)
                with torch.cuda.amp.autocast(enabled=(device.startswith("cuda"))):
                    image_features = model.encode_image(images)
                    text_features  = model.encode_text(tokens)
                    val_loss = clip_loss(image_features, text_features, model.logit_scale)
                val_losses.append(val_loss.item())
            val_mean = sum(val_losses)/len(val_losses)

    ckpt_path = os.path.join(output_dir, f"epoch{epoch}_val{val_mean:.4f}.pt")
    torch.save({"model": model.state_dict()}, ckpt_path)
    if val_mean < best_val:
        best_val = val_mean
        torch.save({"model": model.state_dict()}, os.path.join(output_dir, "best.pt"))

    print(f"Epoch {epoch} done. TrainLoss ~{running/step:.4f}  ValLoss {val_mean:.4f}")        

class FeedbackDataset(Dataset):
    def __init__(self, examples, processor=None):
        self.examples = examples
        self.processor = processor

    def __len__(self):
        return len(self.examples)

    def __getitem__(self, idx):
        ex = self.examples[idx]
        image = ex["image"]
        if not isinstance(image, Image.Image):
            image = Image.open(image).convert("RGB")
        return image, ex["text"], ex["label"]
    
def feedback(model,processor,device,data,epochs=5,batch_size=4,lr=1e-6):
    dataset=FeedbackDataset(data,processor=processor)
    dataLoader=DataLoader(dataset,batch_size=batch_size,shuffle=True)
    optimizer = torch.optim.AdamW(model.parameters(), lr=lr)
    loss_fn = nn.CosineEmbeddingLoss()
    model.load_state_dict(torch.load(model_path, map_location=device))
    model.train()
    for epoch in range(epochs):
        total_loss = 0
        for images, texts, labels in dataLoader:
            inputs = processor(text=texts, images=images,
                               return_tensors="pt", padding=True).to(device)
            text_embeds = model.get_text_features(inputs["input_ids"], inputs["attention_mask"])
            image_embeds = model.get_image_features(inputs["pixel_values"])
            text_embeds = text_embeds / text_embeds.norm(dim=-1, keepdim=True)
            image_embeds = image_embeds / image_embeds.norm(dim=-1, keepdim=True)

            labels = torch.tensor(labels, dtype=torch.float, device=device)
            loss = loss_fn(image_embeds, text_embeds, labels)

            optimizer.zero_grad()
            loss.backward()
            optimizer.step()

            total_loss += loss.item()
        
        print(f"{epoch+1}/{epochs} , Loss :{total_loss/len(dataLoader):.4f}")



def encode_img_and_text(imgs,text):
    image_feat=[]
    model, _, preprocess =open_clip.create_model_and_transforms('ViT-B-32',pretrained='openai',device=device,quick_gelu=True )
    checkpoint = torch.load(model_path, map_location=device)
    model.to(device)
    for img in imgs:
        if hasattr(img, 'read'):
            image = Image.open(img.stream).convert("RGB")
        else:
            if isinstance(img, dict) and 'preview' in img:
                img_data = img['preview'].split(",")[1]
                image = Image.open(BytesIO(base64.b64decode(img_data))).convert("RGB")
            else:
                raise ValueError("Unsupported image input")
        image_input = preprocess(image).unsqueeze(0).to(device)
        with torch.no_grad():
            image_features = model.encode_image(image_input)
            image_features = image_features / image_features.norm(dim=-1, keepdim=True)
            image_feat.append(image_features)
    image_embedding=torch.stack(image_feat).mean(dim=0)
    text_tokens=tokenizer([text]).to(device)
    with torch.no_grad():
        text_features = model.encode_text(text_tokens)
        text_features = text_features / text_features.norm(dim=-1, keepdim=True) 
    alpha=0.7
    combined=alpha*image_embedding+(1-alpha)*text_features
    combined=combined/combined.norm(dim=-1,keepdim=True)
    return combined.squeeze(0).cpu().tolist()