Spaces:
Running
Running
update system prompt
Browse files
app.py
CHANGED
@@ -234,28 +234,35 @@ Functions that typically need @spaces.GPU:
|
|
234 |
|
235 |
## Advanced ZeroGPU Optimization (Recommended)
|
236 |
|
237 |
-
For production Spaces with heavy models,
|
238 |
|
|
|
239 |
```python
|
240 |
import spaces
|
241 |
import torch
|
242 |
from diffusers import DiffusionPipeline
|
243 |
|
244 |
-
|
|
|
245 |
pipe.to('cuda')
|
246 |
|
247 |
-
@spaces.GPU(duration=1500) #
|
248 |
def compile_transformer():
|
|
|
249 |
with spaces.aoti_capture(pipe.transformer) as call:
|
250 |
pipe("arbitrary example prompt")
|
251 |
|
|
|
252 |
exported = torch.export.export(
|
253 |
pipe.transformer,
|
254 |
args=call.args,
|
255 |
kwargs=call.kwargs,
|
256 |
)
|
|
|
|
|
257 |
return spaces.aoti_compile(exported)
|
258 |
|
|
|
259 |
compiled_transformer = compile_transformer()
|
260 |
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
261 |
|
@@ -264,10 +271,163 @@ def generate(prompt):
|
|
264 |
return pipe(prompt).images
|
265 |
```
|
266 |
|
267 |
-
|
268 |
-
|
269 |
-
|
270 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
271 |
|
272 |
## Complete Gradio API Reference
|
273 |
|
@@ -327,28 +487,35 @@ Functions that typically need @spaces.GPU:
|
|
327 |
|
328 |
## Advanced ZeroGPU Optimization (Recommended)
|
329 |
|
330 |
-
For production Spaces with heavy models,
|
331 |
|
|
|
332 |
```python
|
333 |
import spaces
|
334 |
import torch
|
335 |
from diffusers import DiffusionPipeline
|
336 |
|
337 |
-
|
|
|
338 |
pipe.to('cuda')
|
339 |
|
340 |
-
@spaces.GPU(duration=1500) #
|
341 |
def compile_transformer():
|
|
|
342 |
with spaces.aoti_capture(pipe.transformer) as call:
|
343 |
pipe("arbitrary example prompt")
|
344 |
|
|
|
345 |
exported = torch.export.export(
|
346 |
pipe.transformer,
|
347 |
args=call.args,
|
348 |
kwargs=call.kwargs,
|
349 |
)
|
|
|
|
|
350 |
return spaces.aoti_compile(exported)
|
351 |
|
|
|
352 |
compiled_transformer = compile_transformer()
|
353 |
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
354 |
|
@@ -357,10 +524,163 @@ def generate(prompt):
|
|
357 |
return pipe(prompt).images
|
358 |
```
|
359 |
|
360 |
-
|
361 |
-
|
362 |
-
|
363 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
364 |
|
365 |
## Complete Gradio API Reference
|
366 |
|
|
|
234 |
|
235 |
## Advanced ZeroGPU Optimization (Recommended)
|
236 |
|
237 |
+
For production Spaces with heavy models, use ahead-of-time (AoT) compilation for 1.3x-1.8x speedups:
|
238 |
|
239 |
+
### Basic AoT Compilation
|
240 |
```python
|
241 |
import spaces
|
242 |
import torch
|
243 |
from diffusers import DiffusionPipeline
|
244 |
|
245 |
+
MODEL_ID = 'black-forest-labs/FLUX.1-dev'
|
246 |
+
pipe = DiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16)
|
247 |
pipe.to('cuda')
|
248 |
|
249 |
+
@spaces.GPU(duration=1500) # Maximum duration allowed during startup
|
250 |
def compile_transformer():
|
251 |
+
# 1. Capture example inputs
|
252 |
with spaces.aoti_capture(pipe.transformer) as call:
|
253 |
pipe("arbitrary example prompt")
|
254 |
|
255 |
+
# 2. Export the model
|
256 |
exported = torch.export.export(
|
257 |
pipe.transformer,
|
258 |
args=call.args,
|
259 |
kwargs=call.kwargs,
|
260 |
)
|
261 |
+
|
262 |
+
# 3. Compile the exported model
|
263 |
return spaces.aoti_compile(exported)
|
264 |
|
265 |
+
# 4. Apply compiled model to pipeline
|
266 |
compiled_transformer = compile_transformer()
|
267 |
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
268 |
|
|
|
271 |
return pipe(prompt).images
|
272 |
```
|
273 |
|
274 |
+
### Advanced Optimizations
|
275 |
+
|
276 |
+
#### FP8 Quantization (Additional 1.2x speedup on H200)
|
277 |
+
```python
|
278 |
+
from torchao.quantization import quantize_, Float8DynamicActivationFloat8WeightConfig
|
279 |
+
|
280 |
+
@spaces.GPU(duration=1500)
|
281 |
+
def compile_transformer_with_quantization():
|
282 |
+
# Quantize before export for FP8 speedup
|
283 |
+
quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig())
|
284 |
+
|
285 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
286 |
+
pipe("arbitrary example prompt")
|
287 |
+
|
288 |
+
exported = torch.export.export(
|
289 |
+
pipe.transformer,
|
290 |
+
args=call.args,
|
291 |
+
kwargs=call.kwargs,
|
292 |
+
)
|
293 |
+
return spaces.aoti_compile(exported)
|
294 |
+
```
|
295 |
+
|
296 |
+
#### Dynamic Shapes (Variable input sizes)
|
297 |
+
```python
|
298 |
+
from torch.utils._pytree import tree_map
|
299 |
+
|
300 |
+
@spaces.GPU(duration=1500)
|
301 |
+
def compile_transformer_dynamic():
|
302 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
303 |
+
pipe("arbitrary example prompt")
|
304 |
+
|
305 |
+
# Define dynamic dimension ranges (model-dependent)
|
306 |
+
transformer_hidden_dim = torch.export.Dim('hidden', min=4096, max=8212)
|
307 |
+
|
308 |
+
# Map argument names to dynamic dimensions
|
309 |
+
transformer_dynamic_shapes = {
|
310 |
+
"hidden_states": {1: transformer_hidden_dim},
|
311 |
+
"img_ids": {0: transformer_hidden_dim},
|
312 |
+
}
|
313 |
+
|
314 |
+
# Create dynamic shapes structure
|
315 |
+
dynamic_shapes = tree_map(lambda v: None, call.kwargs)
|
316 |
+
dynamic_shapes.update(transformer_dynamic_shapes)
|
317 |
+
|
318 |
+
exported = torch.export.export(
|
319 |
+
pipe.transformer,
|
320 |
+
args=call.args,
|
321 |
+
kwargs=call.kwargs,
|
322 |
+
dynamic_shapes=dynamic_shapes,
|
323 |
+
)
|
324 |
+
return spaces.aoti_compile(exported)
|
325 |
+
```
|
326 |
+
|
327 |
+
#### Multi-Compile for Different Resolutions
|
328 |
+
```python
|
329 |
+
@spaces.GPU(duration=1500)
|
330 |
+
def compile_multiple_resolutions():
|
331 |
+
compiled_models = {}
|
332 |
+
resolutions = [(512, 512), (768, 768), (1024, 1024)]
|
333 |
+
|
334 |
+
for width, height in resolutions:
|
335 |
+
# Capture inputs for specific resolution
|
336 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
337 |
+
pipe(f"test prompt {width}x{height}", width=width, height=height)
|
338 |
+
|
339 |
+
exported = torch.export.export(
|
340 |
+
pipe.transformer,
|
341 |
+
args=call.args,
|
342 |
+
kwargs=call.kwargs,
|
343 |
+
)
|
344 |
+
compiled_models[f"{width}x{height}"] = spaces.aoti_compile(exported)
|
345 |
+
|
346 |
+
return compiled_models
|
347 |
+
|
348 |
+
# Usage with resolution dispatch
|
349 |
+
compiled_models = compile_multiple_resolutions()
|
350 |
+
|
351 |
+
@spaces.GPU
|
352 |
+
def generate_with_resolution(prompt, width=1024, height=1024):
|
353 |
+
resolution_key = f"{width}x{height}"
|
354 |
+
if resolution_key in compiled_models:
|
355 |
+
# Temporarily apply the right compiled model
|
356 |
+
spaces.aoti_apply(compiled_models[resolution_key], pipe.transformer)
|
357 |
+
return pipe(prompt, width=width, height=height).images
|
358 |
+
```
|
359 |
+
|
360 |
+
#### FlashAttention-3 Integration
|
361 |
+
```python
|
362 |
+
from kernels import get_kernel
|
363 |
+
|
364 |
+
# Load pre-built FA3 kernel compatible with H200
|
365 |
+
try:
|
366 |
+
vllm_flash_attn3 = get_kernel("kernels-community/vllm-flash-attn3")
|
367 |
+
print("✅ FlashAttention-3 kernel loaded successfully")
|
368 |
+
except Exception as e:
|
369 |
+
print(f"⚠️ FlashAttention-3 not available: {e}")
|
370 |
+
|
371 |
+
# Custom attention processor example
|
372 |
+
class FlashAttention3Processor:
|
373 |
+
def __call__(self, attn, hidden_states, encoder_hidden_states=None, attention_mask=None):
|
374 |
+
# Use FA3 kernel for attention computation
|
375 |
+
return vllm_flash_attn3(hidden_states, encoder_hidden_states, attention_mask)
|
376 |
+
|
377 |
+
# Apply FA3 processor to model
|
378 |
+
if 'vllm_flash_attn3' in locals():
|
379 |
+
for name, module in pipe.transformer.named_modules():
|
380 |
+
if hasattr(module, 'processor'):
|
381 |
+
module.processor = FlashAttention3Processor()
|
382 |
+
```
|
383 |
+
|
384 |
+
### Complete Optimized Example
|
385 |
+
```python
|
386 |
+
import spaces
|
387 |
+
import torch
|
388 |
+
from diffusers import DiffusionPipeline
|
389 |
+
from torchao.quantization import quantize_, Float8DynamicActivationFloat8WeightConfig
|
390 |
+
|
391 |
+
MODEL_ID = 'black-forest-labs/FLUX.1-dev'
|
392 |
+
pipe = DiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16)
|
393 |
+
pipe.to('cuda')
|
394 |
+
|
395 |
+
@spaces.GPU(duration=1500)
|
396 |
+
def compile_optimized_transformer():
|
397 |
+
# Apply FP8 quantization
|
398 |
+
quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig())
|
399 |
+
|
400 |
+
# Capture inputs
|
401 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
402 |
+
pipe("optimization test prompt")
|
403 |
+
|
404 |
+
# Export and compile
|
405 |
+
exported = torch.export.export(
|
406 |
+
pipe.transformer,
|
407 |
+
args=call.args,
|
408 |
+
kwargs=call.kwargs,
|
409 |
+
)
|
410 |
+
return spaces.aoti_compile(exported)
|
411 |
+
|
412 |
+
# Compile during startup
|
413 |
+
compiled_transformer = compile_optimized_transformer()
|
414 |
+
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
415 |
+
|
416 |
+
@spaces.GPU
|
417 |
+
def generate(prompt):
|
418 |
+
return pipe(prompt).images
|
419 |
+
```
|
420 |
+
|
421 |
+
**Expected Performance Gains:**
|
422 |
+
- Basic AoT: 1.3x-1.8x speedup
|
423 |
+
- + FP8 Quantization: Additional 1.2x speedup
|
424 |
+
- + FlashAttention-3: Additional attention speedup
|
425 |
+
- Total potential: 2x-3x faster inference
|
426 |
+
|
427 |
+
**Hardware Requirements:**
|
428 |
+
- FP8 quantization requires CUDA compute capability ≥ 9.0 (H200 ✅)
|
429 |
+
- FlashAttention-3 works on H200 hardware via kernels library
|
430 |
+
- Dynamic shapes add flexibility for variable input sizes
|
431 |
|
432 |
## Complete Gradio API Reference
|
433 |
|
|
|
487 |
|
488 |
## Advanced ZeroGPU Optimization (Recommended)
|
489 |
|
490 |
+
For production Spaces with heavy models, use ahead-of-time (AoT) compilation for 1.3x-1.8x speedups:
|
491 |
|
492 |
+
### Basic AoT Compilation
|
493 |
```python
|
494 |
import spaces
|
495 |
import torch
|
496 |
from diffusers import DiffusionPipeline
|
497 |
|
498 |
+
MODEL_ID = 'black-forest-labs/FLUX.1-dev'
|
499 |
+
pipe = DiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16)
|
500 |
pipe.to('cuda')
|
501 |
|
502 |
+
@spaces.GPU(duration=1500) # Maximum duration allowed during startup
|
503 |
def compile_transformer():
|
504 |
+
# 1. Capture example inputs
|
505 |
with spaces.aoti_capture(pipe.transformer) as call:
|
506 |
pipe("arbitrary example prompt")
|
507 |
|
508 |
+
# 2. Export the model
|
509 |
exported = torch.export.export(
|
510 |
pipe.transformer,
|
511 |
args=call.args,
|
512 |
kwargs=call.kwargs,
|
513 |
)
|
514 |
+
|
515 |
+
# 3. Compile the exported model
|
516 |
return spaces.aoti_compile(exported)
|
517 |
|
518 |
+
# 4. Apply compiled model to pipeline
|
519 |
compiled_transformer = compile_transformer()
|
520 |
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
521 |
|
|
|
524 |
return pipe(prompt).images
|
525 |
```
|
526 |
|
527 |
+
### Advanced Optimizations
|
528 |
+
|
529 |
+
#### FP8 Quantization (Additional 1.2x speedup on H200)
|
530 |
+
```python
|
531 |
+
from torchao.quantization import quantize_, Float8DynamicActivationFloat8WeightConfig
|
532 |
+
|
533 |
+
@spaces.GPU(duration=1500)
|
534 |
+
def compile_transformer_with_quantization():
|
535 |
+
# Quantize before export for FP8 speedup
|
536 |
+
quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig())
|
537 |
+
|
538 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
539 |
+
pipe("arbitrary example prompt")
|
540 |
+
|
541 |
+
exported = torch.export.export(
|
542 |
+
pipe.transformer,
|
543 |
+
args=call.args,
|
544 |
+
kwargs=call.kwargs,
|
545 |
+
)
|
546 |
+
return spaces.aoti_compile(exported)
|
547 |
+
```
|
548 |
+
|
549 |
+
#### Dynamic Shapes (Variable input sizes)
|
550 |
+
```python
|
551 |
+
from torch.utils._pytree import tree_map
|
552 |
+
|
553 |
+
@spaces.GPU(duration=1500)
|
554 |
+
def compile_transformer_dynamic():
|
555 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
556 |
+
pipe("arbitrary example prompt")
|
557 |
+
|
558 |
+
# Define dynamic dimension ranges (model-dependent)
|
559 |
+
transformer_hidden_dim = torch.export.Dim('hidden', min=4096, max=8212)
|
560 |
+
|
561 |
+
# Map argument names to dynamic dimensions
|
562 |
+
transformer_dynamic_shapes = {
|
563 |
+
"hidden_states": {1: transformer_hidden_dim},
|
564 |
+
"img_ids": {0: transformer_hidden_dim},
|
565 |
+
}
|
566 |
+
|
567 |
+
# Create dynamic shapes structure
|
568 |
+
dynamic_shapes = tree_map(lambda v: None, call.kwargs)
|
569 |
+
dynamic_shapes.update(transformer_dynamic_shapes)
|
570 |
+
|
571 |
+
exported = torch.export.export(
|
572 |
+
pipe.transformer,
|
573 |
+
args=call.args,
|
574 |
+
kwargs=call.kwargs,
|
575 |
+
dynamic_shapes=dynamic_shapes,
|
576 |
+
)
|
577 |
+
return spaces.aoti_compile(exported)
|
578 |
+
```
|
579 |
+
|
580 |
+
#### Multi-Compile for Different Resolutions
|
581 |
+
```python
|
582 |
+
@spaces.GPU(duration=1500)
|
583 |
+
def compile_multiple_resolutions():
|
584 |
+
compiled_models = {}
|
585 |
+
resolutions = [(512, 512), (768, 768), (1024, 1024)]
|
586 |
+
|
587 |
+
for width, height in resolutions:
|
588 |
+
# Capture inputs for specific resolution
|
589 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
590 |
+
pipe(f"test prompt {width}x{height}", width=width, height=height)
|
591 |
+
|
592 |
+
exported = torch.export.export(
|
593 |
+
pipe.transformer,
|
594 |
+
args=call.args,
|
595 |
+
kwargs=call.kwargs,
|
596 |
+
)
|
597 |
+
compiled_models[f"{width}x{height}"] = spaces.aoti_compile(exported)
|
598 |
+
|
599 |
+
return compiled_models
|
600 |
+
|
601 |
+
# Usage with resolution dispatch
|
602 |
+
compiled_models = compile_multiple_resolutions()
|
603 |
+
|
604 |
+
@spaces.GPU
|
605 |
+
def generate_with_resolution(prompt, width=1024, height=1024):
|
606 |
+
resolution_key = f"{width}x{height}"
|
607 |
+
if resolution_key in compiled_models:
|
608 |
+
# Temporarily apply the right compiled model
|
609 |
+
spaces.aoti_apply(compiled_models[resolution_key], pipe.transformer)
|
610 |
+
return pipe(prompt, width=width, height=height).images
|
611 |
+
```
|
612 |
+
|
613 |
+
#### FlashAttention-3 Integration
|
614 |
+
```python
|
615 |
+
from kernels import get_kernel
|
616 |
+
|
617 |
+
# Load pre-built FA3 kernel compatible with H200
|
618 |
+
try:
|
619 |
+
vllm_flash_attn3 = get_kernel("kernels-community/vllm-flash-attn3")
|
620 |
+
print("✅ FlashAttention-3 kernel loaded successfully")
|
621 |
+
except Exception as e:
|
622 |
+
print(f"⚠️ FlashAttention-3 not available: {e}")
|
623 |
+
|
624 |
+
# Custom attention processor example
|
625 |
+
class FlashAttention3Processor:
|
626 |
+
def __call__(self, attn, hidden_states, encoder_hidden_states=None, attention_mask=None):
|
627 |
+
# Use FA3 kernel for attention computation
|
628 |
+
return vllm_flash_attn3(hidden_states, encoder_hidden_states, attention_mask)
|
629 |
+
|
630 |
+
# Apply FA3 processor to model
|
631 |
+
if 'vllm_flash_attn3' in locals():
|
632 |
+
for name, module in pipe.transformer.named_modules():
|
633 |
+
if hasattr(module, 'processor'):
|
634 |
+
module.processor = FlashAttention3Processor()
|
635 |
+
```
|
636 |
+
|
637 |
+
### Complete Optimized Example
|
638 |
+
```python
|
639 |
+
import spaces
|
640 |
+
import torch
|
641 |
+
from diffusers import DiffusionPipeline
|
642 |
+
from torchao.quantization import quantize_, Float8DynamicActivationFloat8WeightConfig
|
643 |
+
|
644 |
+
MODEL_ID = 'black-forest-labs/FLUX.1-dev'
|
645 |
+
pipe = DiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16)
|
646 |
+
pipe.to('cuda')
|
647 |
+
|
648 |
+
@spaces.GPU(duration=1500)
|
649 |
+
def compile_optimized_transformer():
|
650 |
+
# Apply FP8 quantization
|
651 |
+
quantize_(pipe.transformer, Float8DynamicActivationFloat8WeightConfig())
|
652 |
+
|
653 |
+
# Capture inputs
|
654 |
+
with spaces.aoti_capture(pipe.transformer) as call:
|
655 |
+
pipe("optimization test prompt")
|
656 |
+
|
657 |
+
# Export and compile
|
658 |
+
exported = torch.export.export(
|
659 |
+
pipe.transformer,
|
660 |
+
args=call.args,
|
661 |
+
kwargs=call.kwargs,
|
662 |
+
)
|
663 |
+
return spaces.aoti_compile(exported)
|
664 |
+
|
665 |
+
# Compile during startup
|
666 |
+
compiled_transformer = compile_optimized_transformer()
|
667 |
+
spaces.aoti_apply(compiled_transformer, pipe.transformer)
|
668 |
+
|
669 |
+
@spaces.GPU
|
670 |
+
def generate(prompt):
|
671 |
+
return pipe(prompt).images
|
672 |
+
```
|
673 |
+
|
674 |
+
**Expected Performance Gains:**
|
675 |
+
- Basic AoT: 1.3x-1.8x speedup
|
676 |
+
- + FP8 Quantization: Additional 1.2x speedup
|
677 |
+
- + FlashAttention-3: Additional attention speedup
|
678 |
+
- Total potential: 2x-3x faster inference
|
679 |
+
|
680 |
+
**Hardware Requirements:**
|
681 |
+
- FP8 quantization requires CUDA compute capability ≥ 9.0 (H200 ✅)
|
682 |
+
- FlashAttention-3 works on H200 hardware via kernels library
|
683 |
+
- Dynamic shapes add flexibility for variable input sizes
|
684 |
|
685 |
## Complete Gradio API Reference
|
686 |
|