MCPcopy
hub / github.com/deep-floyd/IF / __init__

Method __init__

deepfloyd_if/model/unet.py:353–626  ·  view source on GitHub ↗
(
            self,
            in_channels,
            model_channels,
            out_channels,
            num_res_blocks,
            attention_resolutions,
            activation,
            encoder_dim,
            att_pool_heads,
            encoder_channels,
            image_size,
            disable_self_attentions=None,
            dropout=0,
            channel_mult=(1, 2, 4, 8),
            conv_resample=True,
            dims=2,
            num_classes=None,
            precision='32',
            num_heads=1,
            num_head_channels=-1,
            num_heads_upsample=-1,
            use_scale_shift_norm=False,
            resblock_updown=False,
            efficient_activation=False,
            scale_skip_connection=False,
    )

Source from the content-addressed store, hash-verified

351 """
352
353 def __init__(
354 self,
355 in_channels,
356 model_channels,
357 out_channels,
358 num_res_blocks,
359 attention_resolutions,
360 activation,
361 encoder_dim,
362 att_pool_heads,
363 encoder_channels,
364 image_size,
365 disable_self_attentions=None,
366 dropout=0,
367 channel_mult=(1, 2, 4, 8),
368 conv_resample=True,
369 dims=2,
370 num_classes=None,
371 precision='32',
372 num_heads=1,
373 num_head_channels=-1,
374 num_heads_upsample=-1,
375 use_scale_shift_norm=False,
376 resblock_updown=False,
377 efficient_activation=False,
378 scale_skip_connection=False,
379 ):
380 super().__init__()
381
382 if num_heads_upsample == -1:
383 num_heads_upsample = num_heads
384
385 self.encoder_channels = encoder_channels
386 self.encoder_dim = encoder_dim
387 self.efficient_activation = efficient_activation
388 self.scale_skip_connection = scale_skip_connection
389 self.in_channels = in_channels
390 self.model_channels = model_channels
391 self.out_channels = out_channels
392 self.dropout = dropout
393
394 # adapt attention resolutions
395 if isinstance(attention_resolutions, str):
396 self.attention_resolutions = []
397 for res in attention_resolutions.split(','):
398 self.attention_resolutions.append(image_size // int(res))
399 else:
400 self.attention_resolutions = attention_resolutions
401 self.attention_resolutions = tuple(self.attention_resolutions)
402 #
403
404 # adapt disable self attention resolutions
405 if not disable_self_attentions:
406 self.disable_self_attentions = []
407 elif disable_self_attentions is True:
408 self.disable_self_attentions = attention_resolutions
409 elif isinstance(disable_self_attentions, str):
410 self.disable_self_attentions = []

Callers

nothing calls this directly

Calls 12

linearFunction · 0.85
get_activationFunction · 0.85
conv_ndFunction · 0.85
ResBlockClass · 0.85
AttentionBlockClass · 0.85
DownsampleClass · 0.85
UpsampleClass · 0.85
normalizationFunction · 0.85
zero_moduleFunction · 0.85
AttentionPoolingClass · 0.85
__init__Method · 0.45

Tested by

no test coverage detected