File size: 33,775 Bytes
e041d24
0621f21
 
 
 
 
e041d24
0621f21
 
 
 
 
 
12437ad
0621f21
 
 
 
 
e041d24
9078162
 
c1e0817
 
0621f21
e041d24
0621f21
12437ad
0621f21
 
 
 
 
 
 
 
 
 
 
 
 
 
12437ad
0621f21
 
 
 
 
 
12437ad
0621f21
 
 
 
 
 
 
 
 
 
12437ad
0621f21
 
 
 
 
12437ad
0621f21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e041d24
0621f21
e041d24
952fb23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e041d24
 
 
 
 
 
 
952fb23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0621f21
 
 
 
e041d24
12437ad
 
 
e041d24
 
12437ad
e041d24
12437ad
e041d24
12437ad
e041d24
 
12437ad
e041d24
 
 
12437ad
 
e041d24
12437ad
e041d24
12437ad
 
0621f21
 
3487426
0621f21
e041d24
 
0621f21
e041d24
0621f21
 
 
e041d24
12437ad
0621f21
 
 
 
 
 
 
e041d24
 
0621f21
 
 
 
 
 
952fb23
0621f21
952fb23
 
 
 
12437ad
0621f21
952fb23
 
0621f21
 
 
 
 
 
952fb23
0621f21
 
 
 
 
 
952fb23
0621f21
952fb23
 
 
 
 
 
 
 
 
 
 
 
 
 
c1e0817
e041d24
0621f21
 
952fb23
 
 
 
 
 
 
e041d24
0621f21
952fb23
 
12437ad
0621f21
e041d24
0621f21
c1e0817
0621f21
 
c1e0817
0621f21
 
 
 
 
e041d24
 
0621f21
 
 
 
 
 
 
 
 
 
 
c1e0817
0621f21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c1e0817
0621f21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e041d24
952fb23
e041d24
 
0621f21
 
 
952fb23
 
0621f21
 
 
e041d24
0621f21
 
 
 
e041d24
 
 
 
952fb23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e041d24
 
 
 
0621f21
c1e0817
0621f21
 
e041d24
0621f21
 
 
e041d24
0621f21
 
 
 
 
 
e041d24
0621f21
e041d24
0621f21
e041d24
 
0621f21
 
952fb23
0621f21
 
952fb23
 
 
 
 
 
 
 
 
 
 
 
0621f21
e041d24
952fb23
 
e041d24
952fb23
 
 
0621f21
 
 
 
e041d24
0621f21
e041d24
 
952fb23
e041d24
952fb23
e041d24
 
 
 
0e38dc9
e041d24
 
 
 
 
 
 
 
 
 
 
 
952fb23
 
 
 
 
 
 
e041d24
 
4c32dbb
e041d24
 
 
 
 
 
 
952fb23
e041d24
 
 
 
 
 
 
 
 
 
 
952fb23
e041d24
 
 
 
 
4c32dbb
 
f6227a4
e041d24
 
 
 
 
952fb23
e041d24
 
c1e0817
e041d24
 
 
0621f21
 
 
 
c1e0817
952fb23
0621f21
952fb23
 
e041d24
0621f21
 
e041d24
 
 
 
 
0621f21
c1e0817
e041d24
 
 
 
0621f21
 
e041d24
0621f21
 
3487426
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
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
# app.py
import torch
import torch.nn.functional as F
import torch.distributions as dists
import transformers
from transformers import AutoTokenizer
from peft import PeftModel, PeftConfig
import numpy as np
import random
import time
import os
from typing import List, Dict, Optional, Tuple, Iterator, Set
import gradio as gr
import spaces  # ๅฏผๅ…ฅ spaces ๆจกๅ—

# Suppress some Hugging Face warnings
os.environ["TOKENIZERS_PARALLELISM"] = "false"

# Import necessary model classes from the local directory
# Make sure the 'model_cache' directory is in your Hugging Face Space repository.
from model_cache.llada.modeling_llada import LLaDAModelLM
from model_cache.llada.configuration_llada import LLaDAConfig

# --- Helper Functions (Unchanged) ---
def set_seed(seed):
    torch.manual_seed(seed); random.seed(seed); np.random.seed(seed); 
    if torch.cuda.is_available(): torch.cuda.manual_seed_all(seed); torch.backends.cudnn.deterministic = True; torch.backends.cudnn.benchmark = False

def create_full_block_attention_mask(prompt_length, max_length, block_size, device=None, dtype=None):
    if dtype is None: dtype = torch.bfloat16
    attention_mask = torch.full((1, 1, max_length, max_length), -torch.inf, device=device, dtype=dtype)
    attention_mask[:, :, :prompt_length, :prompt_length] = 0
    remaining_length = max_length - prompt_length
    num_blocks = (remaining_length + block_size - 1) // block_size
    for b in range(num_blocks):
        block_start = prompt_length + b * block_size; block_end = min(prompt_length + (b + 1) * block_size, max_length)
        attention_mask[:, :, block_start:block_end, :prompt_length] = 0
        for prev_b in range(b):
            prev_start = prompt_length + prev_b * block_size; prev_end = min(prompt_length + (prev_b + 1) * block_size, max_length)
            attention_mask[:, :, block_start:block_end, prev_start:prev_end] = 0
        attention_mask[:, :, block_start:block_end, block_start:block_end] = 0
    return attention_mask

def extract_attention_mask(full_mask, start_pos, input_length, cache_length):
    end_pos = start_pos + input_length; total_length = cache_length + input_length
    extracted_mask = torch.full((1, 1, input_length, total_length), -torch.inf, device=full_mask.device, dtype=full_mask.dtype)
    extracted_mask[:, :, :, :cache_length] = full_mask[:, :, start_pos:end_pos, :cache_length]
    extracted_mask[:, :, :, cache_length:] = full_mask[:, :, start_pos:end_pos, start_pos:end_pos]
    return extracted_mask

def top_p_logits(logits, top_p=None):
    sorted_logits, sorted_indices = torch.sort(logits, descending=True)
    cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
    sorted_indices_to_remove = cumulative_probs > top_p
    sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
    sorted_indices_to_remove[..., 0] = 0
    mask = torch.zeros_like(logits, dtype=torch.bool, device=logits.device)
    mask = mask.scatter_(-1, sorted_indices, sorted_indices_to_remove)
    logits = logits.masked_fill(mask, torch.finfo(logits.dtype).min)
    return logits

def top_k_logits(logits, top_k=None):
    top_k = min(top_k, logits.size(-1))
    indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
    logits = logits.masked_fill(indices_to_remove, torch.finfo(logits.dtype).min)
    return logits

def sample_tokens(logits, temperature=0.0, top_p=None, top_k=None, margin_confidence=False, neg_entropy=False):
    if temperature > 0: logits = logits / temperature
    if top_p is not None and top_p < 1: logits = top_p_logits(logits, top_p)
    if top_k is not None: logits = top_k_logits(logits, top_k)
    probs = torch.softmax(logits, dim=-1)
    if temperature > 0:
        try:
            x0 = dists.Categorical(probs=probs).sample()
            initial_confidence = torch.gather(probs, -1, x0.unsqueeze(-1)).squeeze(-1)
        except: initial_confidence, x0 = probs.max(dim=-1)
    else: initial_confidence, x0 = probs.max(dim=-1)
    confidence = initial_confidence.clone()
    if margin_confidence:
        sorted_probs, _ = torch.sort(probs, dim=-1, descending=True)
        confidence = sorted_probs[:, 0] - sorted_probs[:, 1]
    if neg_entropy:
        epsilon = 1e-10
        confidence = torch.sum(probs * torch.log(probs + epsilon), dim=-1)
    return confidence, x0, initial_confidence


class DreamLoRAInference:
    # CSS is exactly the same as your original script
    CSS = """
    /* Enhanced modern styling */
    .main-container {
        max-width: 1400px;
        margin: 0 auto;
        padding: 20px;
    }
    .output-text-container {
        background: linear-gradient(135deg, #f8fafc 0%, #f1f5f9 100%);
        border: 2px solid #e2e8f0;
        border-radius: 12px;
        padding: 20px;
        margin: 15px 0;
        box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
    }
    .output-text-container textarea {
        background: white !important;
        border: 1px solid #cbd5e1 !important;
        border-radius: 8px !important;
        font-family: 'Inter', 'Segoe UI', sans-serif !important;
        font-size: 14px !important;
        line-height: 1.6 !important;
        padding: 16px !important;
        box-shadow: inset 0 2px 4px 0 rgba(0, 0, 0, 0.06) !important;
    }
    .stats-card {
        background: linear-gradient(135deg, #ecfdf5 0%, #f0fdf4 100%);
        border: 2px solid #10b981;
        border-radius: 12px;
        padding: 20px;
        margin: 15px 0;
        box-shadow: 0 4px 6px -1px rgba(16, 185, 129, 0.1);
    }
    .stats-card h3 {
        color: #065f46;
        margin-top: 0;
        margin-bottom: 15px;
        font-weight: 600;
    }
    .stats-grid {
        display: grid;
        grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
        gap: 12px;
        margin-top: 10px;
    }
    .stat-item {
        background: white;
        padding: 12px 16px;
        border-radius: 8px;
        border-left: 4px solid #10b981;
        box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1);
    }
    .stat-label {
        font-size: 12px;
        color: #6b7280;
        text-transform: uppercase;
        letter-spacing: 0.5px;
        margin-bottom: 4px;
    }
    .stat-value {
        font-size: 18px;
        font-weight: 600;
        color: #065f46;
        font-family: 'Monaco', 'Menlo', monospace;
    }
    .viz-container {
        background: linear-gradient(135deg, #fefefe 0%, #f9fafb 100%);
        border: 2px solid #e5e7eb;
        border-radius: 12px;
        padding: 20px;
        margin: 15px 0;
        height: 600px;
        overflow-y: auto;
        box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
        position: relative;
    }
    .viz-header {
        background: linear-gradient(135deg, #3b82f6 0%, #2563eb 100%);
        color: white;
        padding: 12px 20px;
        margin: -20px -20px 20px -20px;
        border-radius: 12px 12px 0 0;
        font-weight: 600;
        font-size: 16px;
        display: flex;
        align-items: center;
        gap: 8px;
    }
    .viz-header::before {
        content: "๐ŸŽฌ";
        font-size: 18px;
    }
    .block-container {
        display: inline-block;
        border: 2px solid transparent;
        border-radius: 10px;
        padding: 8px;
        margin: 6px 2px;
        transition: all 0.3s cubic-bezier(0.4, 0, 0.2, 1);
        background: rgba(255, 255, 255, 0.8);
        backdrop-filter: blur(10px);
    }
    .block-updating {
        border-color: #ff4500 !important;
        box-shadow: 0 0 20px rgba(255, 69, 0, 0.4);
        transform: scale(1.02);
        background: rgba(255, 245, 238, 0.9) !important;
    }
    .token {
        padding: 4px 8px;
        margin: 2px;
        border-radius: 6px;
        display: inline-block;
        line-height: 1.5;
        font-family: 'Monaco', 'Menlo', monospace;
        font-size: 13px;
        font-weight: 500;
        transition: all 0.2s ease;
    }
    .token:hover {
        transform: translateY(-1px);
        box-shadow: 0 2px 4px rgba(0, 0, 0, 0.15);
    }
    .token.prompt { background: linear-gradient(135deg, #f1f5f9 0%, #e2e8f0 100%); color: #475569; border: 1px solid #cbd5e1; }
    .token.gen-0 { background: linear-gradient(135deg, #dbeafe 0%, #bfdbfe 100%); color: #1e40af; border: 1px solid #60a5fa; }
    .token.gen-1 { background: linear-gradient(135deg, #d1fae5 0%, #a7f3d0 100%); color: #065f46; border: 1px solid #34d399; }
    .token.gen-2 { background: linear-gradient(135deg, #fef3c7 0%, #fde68a 100%); color: #92400e; border: 1px solid #fbbf24; }
    .token.gen-3 { background: linear-gradient(135deg, #fee2e2 0%, #fecaca 100%); color: #991b1b; border: 1px solid #f87171; }
    .token.gen-4 { background: linear-gradient(135deg, #e0e7ff 0%, #c7d2fe 100%); color: #3730a3; border: 1px solid #818cf8; }
    .token.gen-5 { background: linear-gradient(135deg, #f3e8ff 0%, #e9d5ff 100%); color: #6b21a8; border: 1px solid #c084fc; }
    .token.mask { 
        background: linear-gradient(135deg, #f9fafb 0%, #f3f4f6 100%); 
        color: #9ca3af; 
        border: 2px dashed #d1d5db; 
        animation: pulse 2s infinite;
    }
    @keyframes pulse {
        0%, 100% { opacity: 1; }
        50% { opacity: 0.6; }
    }
    .control-button {
        background: linear-gradient(135deg, #8b5cf6 0%, #7c3aed 100%) !important;
        border: none !important;
        color: white !important;
        padding: 12px 24px !important;
        border-radius: 10px !important;
        font-weight: 600 !important;
        font-size: 14px !important;
        box-shadow: 0 4px 6px -1px rgba(139, 92, 246, 0.3) !important;
        transition: all 0.3s ease !important;
        display: flex !important;
        align-items: center !important;
        gap: 8px !important;
        margin: 10px 0 !important;
    }
    .control-button:hover {
        transform: translateY(-2px) !important;
        box-shadow: 0 8px 15px -3px rgba(139, 92, 246, 0.4) !important;
    }
    .control-button:active {
        transform: translateY(0) !important;
    }
    .control-button::before {
        content: "๐ŸŽฎ";
        font-size: 16px;
    }
    .param-card {
        background: white;
        border: 1px solid #e5e7eb;
        border-radius: 10px;
        padding: 16px;
        margin: 8px 0;
        box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1);
    }
    .viz-container::-webkit-scrollbar {
        width: 12px;
    }
    .viz-container::-webkit-scrollbar-track {
        background: #f1f5f9;
        border-radius: 6px;
    }
    .viz-container::-webkit-scrollbar-thumb {
        background: linear-gradient(135deg, #94a3b8 0%, #64748b 100%);
        border-radius: 6px;
        border: 2px solid #f1f5f9;
    }
    .viz-container::-webkit-scrollbar-thumb:hover {
        background: linear-gradient(135deg, #64748b 0%, #475569 100%);
    }
    .generating-indicator {
        display: inline-flex;
        align-items: center;
        gap: 8px;
        color: #6366f1;
        font-weight: 500;
    }
    .generating-indicator::after {
        content: "";
        width: 12px;
        height: 12px;
        border: 2px solid #6366f1;
        border-top: 2px solid transparent;
        border-radius: 50%;
        animation: spin 1s linear infinite;
    }
    @keyframes spin {
        0% { transform: rotate(0deg); }
        100% { transform: rotate(360deg); }
    }
    @media (max-width: 768px) {
        .main-container {
            padding: 10px;
        }
        .stats-grid {
            grid-template-columns: 1fr;
        }
        .viz-container {
            height: 400px;
        }
    }
    """

    def __init__(self, **kwargs):
        print("Initializing DreamLoRAInference...")
        # Lazy loading: store config, don't load model yet
        self.config = kwargs
        self.model = None
        self.tokenizer = None

        if self.config.get("dtype") == "bfloat16" and torch.cuda.is_bf16_supported():
            self.target_dtype = torch.bfloat16
        elif self.config.get("dtype") == "float16":
            self.target_dtype = torch.float16
        else:
            self.target_dtype = torch.float32

        # Set attributes from config
        for key, value in kwargs.items():
            setattr(self, key, value)
            
        print("DreamLoRAInference configured. Model will be loaded on first use.")

    def _ensure_model_loaded(self):
        """Load model and tokenizer if they haven't been loaded yet."""
        if self.model is None:
            print("Loading model and tokenizer for the first time...")
            self._setup_model(self.config["pretrained_path"], self.config["lora_path"])
            print("Model and tokenizer setup complete.")

    def _setup_model(self, pretrained_path, lora_path):
        config = LLaDAConfig.from_pretrained(pretrained_path)
        self.model = LLaDAModelLM.from_pretrained(
            pretrained_path, 
            config=config, 
            torch_dtype=self.target_dtype,
            device_map="auto"  # Use device_map for auto hardware assignment
        ).eval()
        self.model = PeftModel.from_pretrained(self.model, lora_path)
        self.tokenizer = AutoTokenizer.from_pretrained(pretrained_path)
        if self.tokenizer.pad_token is None:
            self.tokenizer.pad_token = self.tokenizer.eos_token

    def _apply_chat_template(self, prompt):
        chat_history = [{"role": "user", "content": prompt}]
        return self.tokenizer.apply_chat_template(chat_history, tokenize=False, add_generation_prompt=True)

    def _update_block_completion_states(self, block_states, decoded_token_threshold):
        for block_id in sorted(block_states.keys()):
            if 'total_masks' in block_states[block_id] and block_states[block_id]['total_masks'] > 0:
                decoded_tokens = block_states[block_id]['total_masks'] - block_states[block_id]['mask_count']
                decode_ratio = decoded_tokens / block_states[block_id]['total_masks']
                if decode_ratio >= decoded_token_threshold:
                    if (next_block_id := block_id + 1) in block_states:
                        block_states[next_block_id]['is_complete'] = True

    def _render_visualization_html(self, step: int, x_t: torch.Tensor, block_states: Dict, cache_length: int, updated_block_ids: Set[int]) -> str:
        # This function is identical to your original one, with HTML escaping.
        timestamp = int(time.time() * 1000)
        
        html_parts = []
        html_parts.append('<div class="viz-header">Slow-Motion Generation Process</div>')
        
        for block_id in sorted(k for k in block_states.keys() if k > 0):
            state = block_states[block_id]
            container_classes = ["block-container"]
            if block_id in updated_block_ids: container_classes.append("block-updating")
            html_parts.append(f'<div class="{" ".join(container_classes)}" id="block-{block_id}-{timestamp}">')
            block_tokens = x_t[0, state['start_pos']:state['end_pos']]
            for token_id in block_tokens:
                token_id_int = token_id.item()
                token_classes = ["token"]
                if token_id_int == self.mask_token_id:
                    token_str = 'โ–‘'; token_classes.append("mask")
                else:
                    token_str = self.tokenizer.decode([token_id_int], skip_special_tokens=False)
                    token_str = token_str.replace('&', '&').replace('<', '<').replace('>', '>')
                    token_classes.append(f"gen-{(block_id - 1) % 6}")
                html_parts.append(f'<span class="{" ".join(token_classes)}">{token_str}</span>')
            html_parts.append('</div>')
        
        html_parts.append(f'<div class="scroll-anchor" id="viz-anchor-{timestamp}"></div>')
        
        # Script part from original for scrolling
        complete_html = f"""
        <div class="viz-content" id="viz-content-{timestamp}">
            {''.join(html_parts)}
        </div>
        <script>
        (function() {{
            const container = document.querySelector('.viz-container');
            if (container) {{ container.scrollTop = container.scrollHeight; }}
        }})();
        </script>
        """
        return complete_html

    @spaces.GPU
    @torch.inference_mode()
    def stream_and_capture_for_gradio(
        self,
        prompt_text: str,
        max_new_tokens: int,
        block_size: int,
        block_add_threshold: float,
        decoded_token_threshold: float,
        skip_threshold: float
    ) -> Iterator[Tuple[str, List[str], str, bool]]:
        
        # This is the core generation algorithm, now identical to your original script
        self._ensure_model_loaded()
        
        start_time = time.time()
        captured_frames: List[str] = []
        
        input_ids = self.tokenizer(self._apply_chat_template(prompt_text), return_tensors="pt").input_ids.to(self.model.device)
        prompt_length = input_ids.shape[1]
        
        full_attention_mask = create_full_block_attention_mask(prompt_length, self.max_length, block_size, self.model.device, self.target_dtype)
        x_t = input_ids
        block_states = {0: {'start_pos': 0, 'end_pos': prompt_length, 'mask_count': 0, 'total_masks': prompt_length, 'state': 'to_cache', 'is_complete': True}}
        past_key_values, current_blocks, step, eos_detected, cache_length = None, 0, 0, False, 0
        
        initial_viz_html = self._render_visualization_html(0, x_t, block_states, 0, set())
        captured_frames.append(initial_viz_html)
        yield "", captured_frames, '<div class="generating-indicator">Initializing generation process...</div>', False

        while True:
            step += 1
            updated_block_ids: Set[int] = set()

            if len(block_states) - 1 < (max_new_tokens // block_size) and not eos_detected:
                last_block_id = max(block_states.keys())
                progress = (block_states[last_block_id]['total_masks'] - block_states[last_block_id]['mask_count']) / block_states[last_block_id]['total_masks'] if block_states[last_block_id]['total_masks'] > 0 else 1.0
                if progress >= block_add_threshold:
                    new_block_id = last_block_id + 1; new_start_pos = x_t.shape[1]
                    if new_start_pos + block_size <= self.max_length:
                        x_t = torch.cat([x_t, torch.full((1, block_size), self.mask_token_id, device=self.model.device, dtype=torch.long)], dim=1)
                        block_states[new_block_id] = {'start_pos': new_start_pos, 'end_pos': new_start_pos + block_size, 'mask_count': block_size, 'total_masks': block_size, 'state': 'active', 'is_complete': False}
                        current_blocks += 1

            self._update_block_completion_states(block_states, decoded_token_threshold)
            if (x_t == self.mask_token_id).sum() == 0 and current_blocks == 0: break

            blocks_to_cache = [bid for bid, state in block_states.items() if state['state'] == 'to_cache']
            update_kvcache = 0
            if blocks_to_cache:
                start_pos, end_pos = block_states[min(blocks_to_cache)]['start_pos'], block_states[max(blocks_to_cache)]['end_pos']
                update_kvcache = end_pos - start_pos; input_seq, process_start_pos = x_t[:, start_pos:], start_pos
            else:
                active_blocks = [bid for bid, state in block_states.items() if state['state'] == 'active' and state['start_pos'] >= cache_length]
                if not active_blocks: break
                start_pos = min(block_states[bid]['start_pos'] for bid in active_blocks); input_seq, process_start_pos = x_t[:, start_pos:], start_pos
            
            if input_seq.shape[1] == 0: break

            attention_mask = extract_attention_mask(full_attention_mask, process_start_pos, input_seq.shape[1], cache_length)
            outputs = self.model(input_seq, attention_bias=attention_mask, past_key_values=past_key_values, use_cache=True, update_kvcache=update_kvcache + cache_length)
            if update_kvcache > 0:
                past_key_values = outputs.past_key_values
                for bid in blocks_to_cache: block_states[bid]['state'] = 'in_cache'

            blocks_to_deactivate = []
            for block_id, state in block_states.items():
                if state['state'] != 'active': continue
                block_mask_locs = (x_t[0, state['start_pos']:state['end_pos']] == self.mask_token_id).nonzero().squeeze(-1)
                if block_mask_locs.numel() == 0:
                    blocks_to_deactivate.append(block_id); continue
                logit_offset = state['start_pos'] - process_start_pos
                block_mask_logits = outputs.logits[:, logit_offset + block_mask_locs, :]
                _, x0, initial_confidence = sample_tokens(block_mask_logits.squeeze(0), self.temperature, self.top_p, self.top_k)
                all_indices = (initial_confidence > skip_threshold).nonzero().squeeze(-1)
                if state['is_complete'] and all_indices.numel() == 0 and block_mask_logits.numel() > 0:
                    all_indices = torch.tensor([torch.argmax(initial_confidence)], device=self.model.device)

                if all_indices.numel() > 0:
                    updated_block_ids.add(block_id)
                    positions_to_update = state['start_pos'] + block_mask_locs[all_indices]
                    x_t[0, positions_to_update] = x0[all_indices]; state['mask_count'] -= all_indices.numel()
                    if self.tokenizer.eos_token_id in x0[all_indices]: eos_detected = True
                if state['mask_count'] == 0: blocks_to_deactivate.append(block_id)
            
            for bid in blocks_to_deactivate:
                if block_states[bid]['state'] == 'active' and all(block_states.get(i, {}).get('state') != 'active' for i in range(bid)):
                    block_states[bid]['state'] = 'to_cache'; current_blocks -= 1
            if update_kvcache > 0: cache_length += update_kvcache
            
            generated_ids = x_t[0, prompt_length:]
            valid_ids = generated_ids[generated_ids != self.mask_token_id]
            live_text = self.tokenizer.decode(valid_ids, skip_special_tokens=True)
            
            current_viz_html = self._render_visualization_html(step, x_t, block_states, cache_length, updated_block_ids)
            captured_frames.append(current_viz_html)
            
            yield live_text, captured_frames, f'<div class="generating-indicator">Generating... Step {step}</div>', False

        total_time = time.time() - start_time
        final_generated_ids = x_t[0, prompt_length:]
        eos_positions = (final_generated_ids == self.tokenizer.eos_token_id).nonzero()
        if eos_positions.numel() > 0:
            final_generated_ids = final_generated_ids[:eos_positions[0, 0] + 1]

        final_text = self.tokenizer.decode(final_generated_ids, skip_special_tokens=True)
        final_viz_html = self._render_visualization_html(step, x_t, block_states, cache_length, set())
        captured_frames.append(final_viz_html)
        
        tokens_incl_eos = len(final_generated_ids)
        tokens_excl_eos = len(final_generated_ids[final_generated_ids != self.tokenizer.eos_token_id])
        
        stats_html = f"""
        <div class="stats-card">
            <h3>โœ… Generation Complete!</h3>
            <div class="stats-grid">
                <div class="stat-item">
                    <div class="stat-label">Total Time</div>
                    <div class="stat-value">{total_time:.2f}s</div>
                </div>
                <div class="stat-item">
                    <div class="stat-label">Tokens (incl. EOS)</div>
                    <div class="stat-value">{tokens_incl_eos}</div>
                </div>
                <div class="stat-item">
                    <div class="stat-label">Tokens (excl. EOS)</div>
                    <div class="stat-value">{tokens_excl_eos}</div>
                </div>
                <div class="stat-item">
                    <div class="stat-label">Tokens/Second</div>
                    <div class="stat-value">{(tokens_incl_eos / total_time):.1f}</div>
                </div>
            </div>
        </div>
        """
        yield final_text, captured_frames, stats_html, True


# --- Gradio UI and Event Handlers ---
if __name__ == "__main__":
    # Use Hugging Face Hub model IDs
    config = {
        "pretrained_path": "GSAI-ML/LLaDA-8B-Instruct",
        "lora_path": "SJTU-Deng-Lab/D2F_LLaDA_Instruct_8B_Lora",
        "dtype": "bfloat16", "max_length": 4096,
        "temperature": 0.0, "top_p": None, "top_k": None, "mask_token_id": 126336,
        "sampling_strategy": "default",
    }
    set_seed(42)
    inference_engine = DreamLoRAInference(**config)
    
    def animate_visualization(html_frames_list: List[str], delay: float) -> Iterator[str]:
        if not html_frames_list:
            yield '<div class="viz-header">No visualization data captured</div>'
            return
        for frame in html_frames_list:
            yield frame
            time.sleep(delay)

    # Simplified auto-scroll JS from your original script
    auto_scroll_js = """
    <script>
    function setupAutoScroll(containerSelector, contentSelector) {
        const container = document.querySelector(containerSelector);
        if (!container) return;

        const observer = new MutationObserver(() => {
            container.scrollTop = container.scrollHeight;
        });

        observer.observe(container, {
            childList: true,
            subtree: true
        });
    }

    document.addEventListener('DOMContentLoaded', () => {
        // Use a timeout to ensure Gradio elements are rendered
        setTimeout(() => {
            setupAutoScroll('#live-text-output', 'textarea');
            setupAutoScroll('.viz-container', '.viz-content');
        }, 1500);
    });
    </script>
    """

    with gr.Blocks(css=DreamLoRAInference.CSS, theme=gr.themes.Soft(), title="D2F-LLaDA Visualization") as demo:
        html_frames_state = gr.State([])
        generation_complete_state = gr.State(False)

        gr.HTML(auto_scroll_js) # Keep the JS injection

        # The entire UI layout is now identical to your original script
        with gr.Column(elem_classes=["main-container"]):
            gr.Markdown("# โœจ D2F: Faster-than-AR Inference for Diffusion LLMs")
            gr.Markdown(
                """
                [GitHub](https://github.com/zhijie-group/Discrete-Diffusion-Forcing) | [๐Ÿ“œ Paper](https://arxiv.org/abs/2508.09192) | [๐ŸŒ Blog Post](https://zhijie-group.github.io/Discrete-Diffusion-Forcing/) | [๐Ÿค— D2F-LLaDA LoRA](https://huggingface.co/SJTU-Deng-Lab/D2F_LLaDA_Instruct_8B_Lora) | [๐Ÿค— D2F-Dream LoRA](https://huggingface.co/SJTU-Deng-Lab/D2F_Dream_Base_7B_Lora)
                """
            )
            gr.Markdown(
                """
                This demo showcases **Discrete Diffusion Forcing (D2F)**, a novel framework that enables Diffusion Language Models (dLLMs) to achieve faster-than-autoregressive inference speeds for the first time. D2F creates an AR-Diffusion hybrid paradigm that combines the efficiency of KV Caching with inter-block parallel decoding.

                The model powering this demo is **LLaDA-Instruct-8B**, fine-tuned with our D2F method. Watch its unique block-wise generation in real-time, then replay the process in slow motion to see how it works!
                """
            )
            
            with gr.Row():
                with gr.Column(scale=2):
                    prompt_input = gr.Textbox(
                        label="๐Ÿค” Enter your question", 
                        placeholder="Ask me anything! Try: 'Explain quantum physics' or 'Write a story about...'", 
                        lines=4,
                        elem_classes=["param-card"]
                    )
                    
                    with gr.Accordion("โš™๏ธ Advanced Settings", open=False):
                        with gr.Row():
                            max_new_tokens_slider = gr.Slider(minimum=64, maximum=2048, value=2048, step=64, label="Max Tokens", info="Maximum number of tokens to generate")
                            block_size_slider = gr.Slider(minimum=16, maximum=128, value=32, step=16, label="Block Size", info="Size of each generation block")
                        with gr.Row():
                            block_add_thresh_slider = gr.Slider(minimum=0.0, maximum=1.0, value=0.1, step=0.05, label="Block Add Threshold", info="When to add new blocks")
                            decoded_token_thresh_slider = gr.Slider(minimum=0.0, maximum=1.0, value=0.5, step=0.05, label="Completion Threshold", info="Block completion criteria")
                        with gr.Row():
                            skip_thresh_slider = gr.Slider(minimum=0.0, maximum=1.0, value=0.9, step=0.01, label="Skip Threshold", info="Token selection threshold")
                            delay_slider = gr.Slider(minimum=0.0, maximum=1.0, value=0.15, step=0.05, label="Playback Speed", info="Slow-motion playback delay (seconds)")
                    
                    generate_button = gr.Button("๐Ÿš€ Generate Text", variant="primary", size="lg")

                with gr.Column(scale=3):
                    with gr.Group(elem_classes=["output-text-container"]):
                        gr.Markdown("### ๐Ÿ“ Generated Text (Real-time)")
                        live_text_output = gr.Textbox(label="", interactive=False, lines=15, show_label=False, placeholder="Generated text will appear here as the AI thinks...", elem_id="live-text-output")
                    stats_output = gr.HTML(elem_id="stats-output")
            
            with gr.Row():
                with gr.Column():
                    slowmo_button = gr.Button("๐ŸŽฌ Watch Slow-Motion Generation Process", variant="secondary", size="lg", elem_classes=["control-button"], visible=False, interactive=False)
                    with gr.Group(elem_classes=["viz-container"], visible=False) as viz_group:
                        visualization_output = gr.HTML(label="")

            # Examples are identical to your original script
            gr.Examples(
                examples=[
                    ["A circular swimming pool has a diameter of 8 meters. Calculate the pool's circumference and area. First, explain the relationship between diameter, radius, circumference, and area of a circle, including the role of ฯ€ in these formulas. Then perform the calculations using ฯ€ โ‰ˆ 3.14159. Next, estimate how much water (in cubic meters) would be needed to fill this pool if it has a uniform depth of 1.5 meters. Finally, calculate how much it would cost to fill this pool if water costs $2.50 per cubic meter. Show all steps and include appropriate units in your answer.", 2048, 32, 0.1, 0.5, 0.9, 0.1],
                    ["A movie theater offers a loyalty card that costs $15 and gives a 15% discount on all tickets. If a regular movie ticket costs $10, how many tickets would you need to buy to make the loyalty card worthwhile? First, explain the concept of a break-even point. Then set up an equation to find when the total cost with the card equals the total cost without the card. Solve this equation step by step, showing all your work. Finally, interpret your answer in the context of the problem.", 2048, 32, 0.1, 0.5, 0.9, 0.1],
                    ["Solve the equation xยฒ - 6x + 8 = 0. First, explain what a quadratic equation is and why it can have up to two solutions. Then solve this equation using three different methods: factoring, completing the square, and the quadratic formula. For each method, explain the mathematical reasoning behind it, show all steps in detail, and discuss when this particular method is most useful. Finally, verify your solutions by substituting them back into the original equation.", 2048, 32, 0.1, 0.55, 0.9, 0.1],
                ],
                inputs=[prompt_input, max_new_tokens_slider, block_size_slider, block_add_thresh_slider, decoded_token_thresh_slider, skip_thresh_slider, delay_slider],
                label="๐Ÿ’ก Try these examples"
            )

        # Event handling is now identical to your original, correct script
        def update_slowmo_button_visibility(is_complete):
            return gr.update(visible=is_complete, interactive=is_complete)
        
        def show_visualization():
            return gr.update(visible=True)

        inputs_list = [
            prompt_input, max_new_tokens_slider, block_size_slider,
            block_add_thresh_slider, decoded_token_thresh_slider, skip_thresh_slider
        ]
        
        # This is the original, correct event chain
        generation_event = generate_button.click(
            fn=lambda: [gr.update(visible=False, interactive=False), gr.update(visible=False), gr.update(value=None), gr.update(value="")],
            outputs=[slowmo_button, viz_group, stats_output, live_text_output]
        ).then(
            fn=inference_engine.stream_and_capture_for_gradio,
            inputs=inputs_list,
            outputs=[live_text_output, html_frames_state, stats_output, generation_complete_state]
        ).then(
            fn=update_slowmo_button_visibility,
            inputs=[generation_complete_state],
            outputs=[slowmo_button]
        )
        
        slowmo_event = slowmo_button.click(
            fn=show_visualization,
            outputs=[viz_group]
        ).then(
            fn=animate_visualization,
            inputs=[html_frames_state, delay_slider],
            outputs=[visualization_output]
        )

    demo.queue().launch()