Spaces:
Build error
Build error
File size: 6,327 Bytes
b6af722 |
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 |
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from hydra.core.config_store import ConfigStore
from cosmos_predict1.diffusion.networks.general_dit_video_conditioned import VideoExtendGeneralDIT
from cosmos_predict1.diffusion.training.utils.peft.lora_config import get_fa_ca_qv_lora_config
from cosmos_predict1.utils.lazy_config import LazyCall as L
from cosmos_predict1.utils.lazy_config import LazyDict
Cosmos_Predict1_Video2World_7B: LazyDict = LazyDict(
dict(
defaults=[
{"override /net": "faditv2_7b"},
{"override /conditioner": "video_cond"},
{"override /tokenizer": "cosmos_diffusion_tokenizer_res720_comp8x8x8_t121_ver092624"},
"_self_",
],
model=dict(
latent_shape=[
16,
16,
88,
160,
],
conditioner=dict(video_cond_bool=dict()),
net=L(VideoExtendGeneralDIT)(
rope_h_extrapolation_ratio=1.0,
rope_w_extrapolation_ratio=1.0,
rope_t_extrapolation_ratio=2.0,
),
),
job=dict(group="Video2World", name="Cosmos_Predict1_Video2World_7B"),
)
)
Cosmos_Predict1_Video2World_14B: LazyDict = LazyDict(
dict(
defaults=[
{"override /net": "faditv2_14b"},
{"override /conditioner": "video_cond"},
{"override /tokenizer": "cosmos_diffusion_tokenizer_res720_comp8x8x8_t121_ver092624"},
"_self_",
],
model=dict(
latent_shape=[
16,
16,
88,
160,
],
conditioner=dict(video_cond_bool=dict()),
net=L(VideoExtendGeneralDIT)(
rope_h_extrapolation_ratio=2.0,
rope_t_extrapolation_ratio=2.0,
rope_w_extrapolation_ratio=2.0,
extra_h_extrapolation_ratio=2.0,
extra_t_extrapolation_ratio=2.0,
extra_w_extrapolation_ratio=2.0,
),
),
job=dict(group="Video2World", name="Cosmos_Predict1_Video2World_14B"),
)
)
Cosmos_Predict1_Video2World_7B_Post_trained: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_7B",
],
job=dict(
name="Cosmos_Predict1_Video2World_7B_Post_trained",
),
)
)
Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_80gb: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_7B",
],
job=dict(
name="Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_80gb",
),
model=dict(
latent_shape=[
16, # Latent channel dim
16, # Latent temporal dim
48, # Latent height dim
48, # Latent width dim
],
tokenizer=dict(
video_vae=dict(pixel_chunk_duration=121, spatial_resolution="384"),
),
),
)
)
Cosmos_Predict1_Video2World_7B_Post_trained_8gpu_40gb: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_7B",
],
job=dict(
name="Cosmos_Predict1_Video2World_7B_Post_trained_8gpu_40gb",
),
model=dict(
latent_shape=[
16, # Latent channel dim
16, # Latent temporal dim
48, # Latent height dim
48, # Latent width dim
],
tokenizer=dict(
video_vae=dict(pixel_chunk_duration=25, spatial_resolution="384"),
),
),
)
)
Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_40gb: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_7B",
],
job=dict(
name="Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_40gb",
),
model=dict(
latent_shape=[
16, # Latent channel dim
16, # Latent temporal dim
24, # Latent height dim
24, # Latent width dim
],
tokenizer=dict(
# video_vae=dict(pixel_chunk_duration=17, spatial_resolution="384"),
video_vae=dict(pixel_chunk_duration=25, spatial_resolution="384"),
),
),
)
)
Cosmos_Predict1_Video2World_14B_Post_trained: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_14B",
],
job=dict(
name="Cosmos_Predict1_Video2World_14B_Post_trained",
),
)
)
Cosmos_Predict1_Video2World_7B_Post_trained_lora: LazyDict = LazyDict(
dict(
defaults=[
"/experiment/Cosmos_Predict1_Video2World_7B_Post_trained",
],
job=dict(
name="Cosmos_Predict1_Video2World_7B_Post_trained_lora",
),
model=dict(
peft_control=get_fa_ca_qv_lora_config(first_nblocks=27, rank=8, scale=1),
),
)
)
cs = ConfigStore.instance()
for _item in [
Cosmos_Predict1_Video2World_7B,
Cosmos_Predict1_Video2World_14B,
Cosmos_Predict1_Video2World_7B_Post_trained,
Cosmos_Predict1_Video2World_14B_Post_trained,
Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_80gb,
Cosmos_Predict1_Video2World_7B_Post_trained_8gpu_40gb,
Cosmos_Predict1_Video2World_7B_Post_trained_4gpu_40gb,
Cosmos_Predict1_Video2World_7B_Post_trained_lora,
]:
cs.store(group="experiment", package="_global_", name=_item["job"]["name"], node=_item)
|