File size: 16,218 Bytes
7f615fd
 
 
 
 
 
 
 
 
 
fdd626d
0d6c60b
 
 
 
 
 
 
9efc461
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2a57dcf
9efc461
 
 
 
 
 
 
 
 
6bcc1a3
9efc461
 
7f615fd
 
 
 
 
 
 
2a57dcf
7f615fd
1b040ff
2a57dcf
9efc461
 
1b040ff
2a57dcf
7f615fd
 
0d6c60b
7f615fd
2a57dcf
1b040ff
9efc461
 
 
 
 
6bcc1a3
2a57dcf
9efc461
2a57dcf
9efc461
 
 
2a57dcf
1b040ff
9efc461
1b040ff
 
2a57dcf
9efc461
 
 
 
 
 
 
 
7f615fd
9efc461
1b040ff
9efc461
2a57dcf
7f615fd
1b040ff
9efc461
2a57dcf
9efc461
1b040ff
2a57dcf
9efc461
 
 
 
 
 
 
2a57dcf
9efc461
 
2a57dcf
 
9efc461
 
 
 
2a57dcf
9efc461
7f615fd
 
9efc461
 
7f615fd
7153add
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1b040ff
0d6c60b
 
 
7153add
0d6c60b
7153add
0d6c60b
7153add
0d6c60b
7153add
 
0d6c60b
 
 
1b040ff
 
 
0d6c60b
1b040ff
0d6c60b
 
 
 
 
9efc461
0d6c60b
 
1b040ff
 
0d6c60b
 
 
 
 
1b040ff
0d6c60b
 
1b040ff
0d6c60b
 
 
 
 
 
9efc461
0d6c60b
 
 
 
 
 
 
7f615fd
1b040ff
 
 
 
 
2a57dcf
0d6c60b
7f615fd
 
1b040ff
0d6c60b
1b040ff
7f615fd
2a57dcf
9efc461
 
 
fdd626d
2a57dcf
7f615fd
 
2a57dcf
1b040ff
0d6c60b
9efc461
7f615fd
2a57dcf
7f615fd
9efc461
6bcc1a3
2a57dcf
7f615fd
 
 
 
 
9efc461
 
7f615fd
9efc461
7f615fd
9efc461
1b040ff
 
 
9efc461
 
6bcc1a3
1b040ff
 
 
 
fdd626d
1b040ff
9efc461
6bcc1a3
9efc461
7f615fd
9efc461
fdd626d
1b040ff
9efc461
2a57dcf
9efc461
 
 
 
 
1b040ff
9efc461
 
 
 
1b040ff
7f615fd
9efc461
 
 
 
 
 
7f615fd
2a57dcf
1b040ff
7f615fd
2a57dcf
0d6c60b
 
1b040ff
0d6c60b
 
 
 
 
2a57dcf
7f615fd
 
 
1b040ff
9efc461
6bcc1a3
9efc461
2a57dcf
7f615fd
9efc461
 
2a57dcf
7f615fd
0d6c60b
 
7f615fd
 
9efc461
 
 
2a57dcf
7f615fd
 
1b040ff
 
 
2a57dcf
9efc461
6bcc1a3
9efc461
6bcc1a3
9efc461
 
 
6bcc1a3
9efc461
 
6bcc1a3
2a57dcf
6bcc1a3
 
 
2a57dcf
7f615fd
1b040ff
9efc461
6bcc1a3
2a57dcf
6bcc1a3
 
2a57dcf
7f615fd
1b040ff
2a57dcf
7f615fd
 
0d6c60b
 
 
 
 
9efc461
0d6c60b
 
 
 
 
 
6bcc1a3
7f615fd
 
2a57dcf
7f615fd
 
9efc461
 
 
7f615fd
9efc461
6bcc1a3
7f615fd
2a57dcf
6bcc1a3
9efc461
 
 
 
 
 
 
 
2a57dcf
9efc461
 
 
 
 
 
 
 
 
 
 
2a57dcf
9efc461
6bcc1a3
7f615fd
 
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
import gradio as gr
import os
import tempfile
import shutil
import re
import json
from pathlib import Path
from huggingface_hub import HfApi, hf_hub_download
from safetensors.torch import load_file, save_file
import torch
import torch.nn.functional as F
try:
    from modelscope.hub.file_download import model_file_download as ms_file_download
    from modelscope.hub.api import HubApi as ModelScopeApi
    MODELScope_AVAILABLE = True
except ImportError:
    MODELScope_AVAILABLE = False

def extract_correction_factors(original_weight, fp8_weight):
    """Extract per-channel/tensor correction factors instead of LoRA decomposition."""
    with torch.no_grad():
        # Convert to float32 for precision
        orig = original_weight.float()
        quant = fp8_weight.float()
        
        # Compute error (what needs to be added to FP8 to recover original)
        error = orig - quant
        
        # Skip if error is negligible
        error_norm = torch.norm(error)
        orig_norm = torch.norm(orig)
        if orig_norm > 1e-6 and error_norm / orig_norm < 0.01:
            return None
            
        # For 2D+ tensors, compute per-channel correction (better than LoRA for quantization error)
        if orig.ndim >= 2:
            # Find channel dimension - typically dim 0 for most layers
            channel_dim = 0
            channel_mean = error.mean(dim=tuple(i for i in range(orig.ndim) if i != channel_dim), keepdim=True)
            return channel_mean.to(original_weight.dtype)
        else:
            # For bias/batchnorm etc., use scalar correction
            return error.mean().to(original_weight.dtype)

def convert_safetensors_to_fp8_with_correction(safetensors_path, output_dir, fp8_format, correction_mode="per_channel", progress=gr.Progress()):
    progress(0.1, desc="Starting FP8 conversion with precision recovery...")
    try:
        def read_safetensors_metadata(path):
            with open(path, 'rb') as f:
                header_size = int.from_bytes(f.read(8), 'little')
                header_json = f.read(header_size).decode('utf-8')
                header = json.loads(header_json)
                return header.get('__metadata__', {})
        
        metadata = read_safetensors_metadata(safetensors_path)
        progress(0.2, desc="Loaded metadata.")
        
        # Load original weights for comparison
        original_state = load_file(safetensors_path)
        progress(0.4, desc="Loaded weights.")
        
        if fp8_format == "e5m2":
            fp8_dtype = torch.float8_e5m2
        else:
            fp8_dtype = torch.float8_e4m3fn
        
        sd_fp8 = {}
        correction_factors = {}
        correction_stats = {
            "total_layers": len(original_state),
            "layers_with_correction": 0,
            "skipped_layers": []
        }
        
        total = len(original_state)
        
        for i, key in enumerate(original_state):
            progress(0.4 + 0.4 * (i / total), desc=f"Processing {i+1}/{total}...")
            weight = original_state[key]
            
            if weight.dtype in [torch.float16, torch.float32, torch.bfloat16]:
                # Convert to FP8
                fp8_weight = weight.to(fp8_dtype)
                sd_fp8[key] = fp8_weight
                
                # Generate correction factors
                if correction_mode != "none":
                    corr = extract_correction_factors(weight, fp8_weight)
                    if corr is not None:
                        correction_factors[f"correction.{key}"] = corr
                        correction_stats["layers_with_correction"] += 1
                    else:
                        correction_stats["skipped_layers"].append(f"{key}: negligible error")
            else:
                # Non-float weights (int, bool, etc.) - keep as is
                sd_fp8[key] = weight
                correction_stats["skipped_layers"].append(f"{key}: non-float dtype")
        
        base_name = os.path.splitext(os.path.basename(safetensors_path))[0]
        fp8_path = os.path.join(output_dir, f"{base_name}-fp8-{fp8_format}.safetensors")
        correction_path = os.path.join(output_dir, f"{base_name}-correction.safetensors")
        
        # Save FP8 model
        save_file(sd_fp8, fp8_path, metadata={"format": "pt", "fp8_format": fp8_format, **metadata})
        
        # Save correction factors if any exist
        if correction_factors:
            save_file(correction_factors, correction_path, metadata={
                "format": "pt",
                "correction_mode": correction_mode,
                "stats": json.dumps(correction_stats)
            })
        
        progress(0.9, desc="Saved FP8 and correction files.")
        progress(1.0, desc="βœ… FP8 conversion with precision recovery complete!")
        
        stats_msg = f"""
πŸ“Š Precision Recovery Statistics:
- Total layers: {correction_stats['total_layers']}
- Layers with correction: {correction_stats['layers_with_correction']}
- Correction mode: {correction_mode}
"""
        return True, f"FP8 ({fp8_format}) with precision recovery saved.\n{stats_msg}", correction_stats

    except Exception as e:
        import traceback
        return False, f"Error: {str(e)}\n{traceback.format_exc()}", None

def parse_hf_url(url):
    url = url.strip().rstrip("/")
    if not url.startswith("https://huggingface.co/"):
        raise ValueError("URL must start with https://huggingface.co/")
    path = url.replace("https://huggingface.co/", "")
    parts = path.split("/")
    if len(parts) < 2:
        raise ValueError("Invalid repo format")
    repo_id = "/".join(parts[:2])
    subfolder = ""
    if len(parts) > 3 and parts[2] == "tree":
        subfolder = "/".join(parts[4:]) if len(parts) > 4 else ""
    elif len(parts) > 2:
        subfolder = "/".join(parts[2:])
    return repo_id, subfolder

def download_safetensors_file(source_type, repo_url, filename, hf_token=None, progress=gr.Progress()):
    temp_dir = tempfile.mkdtemp()
    try:
        if source_type == "huggingface":
            repo_id, subfolder = parse_hf_url(repo_url)
            safetensors_path = hf_hub_download(
                repo_id=repo_id,
                filename=filename,
                subfolder=subfolder or None,
                cache_dir=temp_dir,
                token=hf_token,
                resume_download=True
            )
        elif source_type == "modelscope":
            if not MODELScope_AVAILABLE:
                raise ImportError("ModelScope not installed")
            repo_id = repo_url.strip()
            safetensors_path = ms_file_download(model_id=repo_id, file_path=filename)
        else:
            raise ValueError("Unknown source")
        return safetensors_path, temp_dir
    except Exception as e:
        shutil.rmtree(temp_dir, ignore_errors=True)
        raise e

def upload_to_target(target_type, new_repo_id, output_dir, fp8_format, hf_token=None, modelscope_token=None, private_repo=False):
    if target_type == "huggingface":
        api = HfApi(token=hf_token)
        api.create_repo(repo_id=new_repo_id, private=private_repo, repo_type="model", exist_ok=True)
        api.upload_folder(repo_id=new_repo_id, folder_path=output_dir, repo_type="model", token=hf_token)
        return f"https://huggingface.co/{new_repo_id}"
    elif target_type == "modelscope":
        api = ModelScopeApi()
        if modelscope_token:
            api.login(modelscope_token)
        api.push_model(model_id=new_repo_id, model_dir=output_dir)
        return f"https://modelscope.cn/models/{new_repo_id}"
    else:
        raise ValueError("Unknown target")

def process_and_upload_fp8(
    source_type,
    repo_url,
    safetensors_filename,
    fp8_format,
    correction_mode,
    target_type,
    new_repo_id,
    hf_token,
    modelscope_token,
    private_repo,
    progress=gr.Progress()
):
    if not re.match(r"^[a-zA-Z0-9._-]+/[a-zA-Z0-9._-]+$", new_repo_id):
        return None, "❌ Invalid repo ID format. Use 'username/model-name'.", ""
    if source_type == "huggingface" and not hf_token:
        return None, "❌ Hugging Face token required for source.", ""
    if target_type == "huggingface" and not hf_token:
        return None, "❌ Hugging Face token required for target.", ""
    
    temp_dir = None
    output_dir = tempfile.mkdtemp()
    try:
        progress(0.05, desc="Downloading model...")
        safetensors_path, temp_dir = download_safetensors_file(
            source_type, repo_url, safetensors_filename, hf_token, progress
        )
        
        progress(0.25, desc="Converting to FP8 with precision recovery...")
        success, msg, stats = convert_safetensors_to_fp8_with_correction(
            safetensors_path, output_dir, fp8_format, correction_mode, progress
        )
        
        if not success:
            return None, f"❌ Conversion failed: {msg}", ""
        
        progress(0.9, desc="Uploading...")
        repo_url_final = upload_to_target(
            target_type, new_repo_id, output_dir, fp8_format, hf_token, modelscope_token, private_repo
        )
        
        base_name = os.path.splitext(safetensors_filename)[0]
        correction_filename = f"{base_name}-correction.safetensors"
        fp8_filename = f"{base_name}-fp8-{fp8_format}.safetensors"
        
        readme = f"""---
library_name: diffusers
tags:
- fp8
- safetensors
- quantization
- precision-recovery
- diffusion
- converted-by-gradio
---
# FP8 Model with Precision Recovery
- **Source**: `{repo_url}`
- **File**: `{safetensors_filename}`
- **FP8 Format**: `{fp8_format.upper()}`
- **Correction Mode**: {correction_mode}
- **Correction File**: `{correction_filename}`
- **FP8 File**: `{fp8_filename}`

## Usage (Inference)
```python
from safetensors.torch import load_file
import torch

# Load FP8 model and correction factors
fp8_state = load_file("{fp8_filename}")
correction_state = load_file("{correction_filename}") if os.path.exists("{correction_filename}") else {{}}

# Reconstruct high-precision weights
reconstructed = {{}}
for key in fp8_state:
    fp8_weight = fp8_state[key].to(torch.float32)
    
    # Apply correction if available
    correction_key = f"correction.{{key}}"
    if correction_key in correction_state:
        correction = correction_state[correction_key].to(torch.float32)
        reconstructed[key] = fp8_weight + correction
    else:
        reconstructed[key] = fp8_weight

# Use reconstructed weights in your model
model.load_state_dict(reconstructed)
```

## Correction Modes
- **Per-Channel**: Computes mean correction per output channel (best for most layers)
- **Per-Tensor**: Single correction value per tensor (lightweight)
- **None**: No correction (pure FP8)

> Requires PyTorch β‰₯ 2.1 for FP8 support. For best quality, use the correction file during inference.
"""
        
        with open(os.path.join(output_dir, "README.md"), "w") as f:
            f.write(readme)
        
        if target_type == "huggingface":
            HfApi(token=hf_token).upload_file(
                path_or_fileobj=os.path.join(output_dir, "README.md"),
                path_in_repo="README.md",
                repo_id=new_repo_id,
                repo_type="model",
                token=hf_token
            )
        
        progress(1.0, desc="βœ… Done!")
        result_html = f"""
βœ… Success!  
Model uploaded to: <a href="{repo_url_final}" target="_blank">{new_repo_id}</a>  
Includes: FP8 model + precision recovery corrections.
"""
        return gr.HTML(result_html), "βœ… FP8 conversion with precision recovery successful!", msg
    
    except Exception as e:
        import traceback
        return None, f"❌ Error: {str(e)}\n{traceback.format_exc()}", ""
    
    finally:
        if temp_dir:
            shutil.rmtree(temp_dir, ignore_errors=True)
        shutil.rmtree(output_dir, ignore_errors=True)

with gr.Blocks(title="FP8 Quantizer with Precision Recovery") as demo:
    gr.Markdown("# πŸ”„ FP8 Quantizer with Precision Recovery")
    gr.Markdown("Convert `.safetensors` β†’ **FP8** + **correction factors** to recover quantization precision. Supports Hugging Face ↔ ModelScope.")
    
    with gr.Row():
        with gr.Column():
            source_type = gr.Radio(["huggingface", "modelscope"], value="huggingface", label="Source")
            repo_url = gr.Textbox(label="Repo URL or ID", placeholder="https://huggingface.co/... or modelscope-id")
            safetensors_filename = gr.Textbox(label="Filename", placeholder="model.safetensors")
            
            with gr.Accordion("Quantization Settings", open=True):
                fp8_format = gr.Radio(["e4m3fn", "e5m2"], value="e5m2", label="FP8 Format")
                correction_mode = gr.Dropdown(
                    choices=[
                        ("Per-Channel Correction (recommended)", "per_channel"),
                        ("Per-Tensor Correction", "per_tensor"),
                        ("No Correction (pure FP8)", "none")
                    ],
                    value="per_channel",
                    label="Precision Recovery Mode"
                )
            
            with gr.Accordion("Authentication", open=False):
                hf_token = gr.Textbox(label="Hugging Face Token", type="password")
                modelscope_token = gr.Textbox(label="ModelScope Token (optional)", type="password", visible=MODELScope_AVAILABLE)
        
        with gr.Column():
            target_type = gr.Radio(["huggingface", "modelscope"], value="huggingface", label="Target")
            new_repo_id = gr.Textbox(label="New Repo ID", placeholder="user/model-fp8")
            private_repo = gr.Checkbox(label="Private Repository (HF only)", value=False)
            
            status_output = gr.Markdown()
            detailed_log = gr.Textbox(label="Processing Log", interactive=False, lines=10)
    
    convert_btn = gr.Button("πŸš€ Convert & Upload", variant="primary")
    repo_link_output = gr.HTML()
    
    convert_btn.click(
        fn=process_and_upload_fp8,
        inputs=[
            source_type,
            repo_url,
            safetensors_filename,
            fp8_format,
            correction_mode,
            target_type,
            new_repo_id,
            hf_token,
            modelscope_token,
            private_repo
        ],
        outputs=[repo_link_output, status_output, detailed_log],
        show_progress=True
    )
    
    gr.Examples(
        examples=[
            ["huggingface", "https://huggingface.co/Yabo/FramePainter/tree/main", "unet_diffusion_pytorch_model.safetensors", "e5m2", "per_channel", "huggingface"],
            ["huggingface", "https://huggingface.co/stabilityai/sdxl-vae", "diffusion_pytorch_model.safetensors", "e4m3fn", "per_channel", "huggingface"],
            ["huggingface", "https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main/text_encoder", "model.safetensors", "e5m2", "per_channel", "huggingface"]
        ],
        inputs=[source_type, repo_url, safetensors_filename, fp8_format, correction_mode, target_type],
        label="Example Conversions"
    )
    
    gr.Markdown("""
    ## πŸ’‘ Why This Works Better Than LoRA
    
    Traditional LoRA struggles with quantization errors because:
    - LoRA is designed for *weight updates*, not *quantization error recovery*
    - Per-channel correction captures systematic quantization bias better
    - Simpler math β†’ more reliable reconstruction
    
    ## πŸ“Š Precision Recovery Modes
    
    - **Per-Channel (recommended)**: One correction value per output channel
      - Best quality, moderate file size increase (~5-10%)
      - Handles channel-wise quantization bias effectively
      
    - **Per-Tensor**: One correction value per tensor
      - Good balance of quality and file size
      - Better than no correction for most layers
      
    - **None**: Pure FP8 quantization
      - Smallest file size
      - Lowest quality (use only for memory-constrained deployments)
    
    > **Note**: For diffusion models, per-channel correction typically recovers 95%+ of FP16 quality while keeping 70-80% of FP8's memory savings.
    """)

demo.launch()