Upload folder using huggingface_hub
Browse files- .gitattributes +8 -0
- normal/checkpoints/lora_weights_step_00250.safetensors +3 -0
- normal/checkpoints/lora_weights_step_00500.safetensors +3 -0
- normal/checkpoints/lora_weights_step_00750.safetensors +3 -0
- normal/checkpoints/lora_weights_step_01000.safetensors +3 -0
- normal/samples/step_000250_1.mp4 +3 -0
- normal/samples/step_000250_2.mp4 +3 -0
- normal/samples/step_000500_1.mp4 +3 -0
- normal/samples/step_000500_2.mp4 +3 -0
- normal/samples/step_000750_1.mp4 +3 -0
- normal/samples/step_000750_2.mp4 +3 -0
- normal/samples/step_001000_1.mp4 +3 -0
- normal/samples/step_001000_2.mp4 +3 -0
- normal/training_config.yaml +92 -0
.gitattributes
CHANGED
|
@@ -515,3 +515,11 @@ lion/samples/step_001250_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
|
| 515 |
lion/samples/step_001250_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 516 |
lion/samples/step_001500_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 517 |
lion/samples/step_001500_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 515 |
lion/samples/step_001250_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 516 |
lion/samples/step_001500_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 517 |
lion/samples/step_001500_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 518 |
+
normal/samples/step_000250_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 519 |
+
normal/samples/step_000250_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 520 |
+
normal/samples/step_000500_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 521 |
+
normal/samples/step_000500_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 522 |
+
normal/samples/step_000750_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 523 |
+
normal/samples/step_000750_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 524 |
+
normal/samples/step_001000_1.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 525 |
+
normal/samples/step_001000_2.mp4 filter=lfs diff=lfs merge=lfs -text
|
normal/checkpoints/lora_weights_step_00250.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d92fe1e77d0b0f972eced928953e56f0069b4ec4f0008c785db5b9fdcecf6c99
|
| 3 |
+
size 1316096704
|
normal/checkpoints/lora_weights_step_00500.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:909a4e971d314d859106e43608c160e77ceb199bfeed1d52e3dc896881f6c0ad
|
| 3 |
+
size 1316096704
|
normal/checkpoints/lora_weights_step_00750.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a3605474a16861595238987b6a60a8359df4b3329b7b66393a5f415ec102e576
|
| 3 |
+
size 1316096704
|
normal/checkpoints/lora_weights_step_01000.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c9839f7509bd40e9674b0480bec1136606b3d8b09c23388bf41855cde1d1e856
|
| 3 |
+
size 1316096704
|
normal/samples/step_000250_1.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f40619e39e56ed7f21d884e94d7cac4a2f8192b43f701639ded2418b336aeca
|
| 3 |
+
size 1272207
|
normal/samples/step_000250_2.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f56229b8dc6d0ff1b42d63bf3da31f2a5b1131c056a2ed7385e68f2cc5b1e0e6
|
| 3 |
+
size 570705
|
normal/samples/step_000500_1.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93acce041c015485ddb371f072439f0fd0d0f43e27ed10eba79ada694ccd0c29
|
| 3 |
+
size 386262
|
normal/samples/step_000500_2.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:16077f73a6c244b1a7c84364c6bf8708f09b6e218c70ae1a8b3ceff75c97c2a2
|
| 3 |
+
size 278292
|
normal/samples/step_000750_1.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36509b9546c7ecf64faef5994c68ebeda244fa013bf9ebd081771d4970fca2f7
|
| 3 |
+
size 1406338
|
normal/samples/step_000750_2.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d81f820feaa9a70802cb8da7437aa04c363d0ee5e53897b82ff6a2c9f15d9d4
|
| 3 |
+
size 288277
|
normal/samples/step_001000_1.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0985b84194e69c52cbdcabc906a83f3ca894eaf63ae7a78f570dfd0d41256f17
|
| 3 |
+
size 1338004
|
normal/samples/step_001000_2.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:92e8836c481c627782a637e73978b367201c94fe782e791802d62f9cd3497e0f
|
| 3 |
+
size 213013
|
normal/training_config.yaml
ADDED
|
@@ -0,0 +1,92 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
acceleration:
|
| 2 |
+
load_text_encoder_in_8bit: false
|
| 3 |
+
mixed_precision_mode: bf16
|
| 4 |
+
quantization: null
|
| 5 |
+
checkpoints:
|
| 6 |
+
interval: 250
|
| 7 |
+
keep_last_n: 10
|
| 8 |
+
precision: bfloat16
|
| 9 |
+
data:
|
| 10 |
+
num_dataloader_workers: 2
|
| 11 |
+
preprocessed_data_root: .precomputed
|
| 12 |
+
flow_matching:
|
| 13 |
+
timestep_sampling_mode: shifted_logit_normal
|
| 14 |
+
timestep_sampling_params: {}
|
| 15 |
+
hub:
|
| 16 |
+
hub_model_id: Alissonerdx/BFS-LTX2
|
| 17 |
+
push_to_hub: true
|
| 18 |
+
lora:
|
| 19 |
+
alpha: 128
|
| 20 |
+
dropout: 0.01
|
| 21 |
+
rank: 128
|
| 22 |
+
target_modules:
|
| 23 |
+
- attn1.to_k
|
| 24 |
+
- attn1.to_q
|
| 25 |
+
- attn1.to_v
|
| 26 |
+
- attn1.to_out.0
|
| 27 |
+
- attn2.to_k
|
| 28 |
+
- attn2.to_q
|
| 29 |
+
- attn2.to_v
|
| 30 |
+
- attn2.to_out.0
|
| 31 |
+
- ff.net.0.proj
|
| 32 |
+
- ff.net.2
|
| 33 |
+
model:
|
| 34 |
+
load_checkpoint: /train/LTX-2/packages/ltx-trainer/outputs/ltx2_v2v_ic_lora_head_swap_128_128_normal/checkpoints
|
| 35 |
+
model_path: /train/models/ltx-2-19b-dev.safetensors
|
| 36 |
+
text_encoder_path: /train/models/gemma
|
| 37 |
+
training_mode: lora
|
| 38 |
+
optimization:
|
| 39 |
+
batch_size: 1
|
| 40 |
+
enable_gradient_checkpointing: true
|
| 41 |
+
gradient_accumulation_steps: 1
|
| 42 |
+
learning_rate: 1.0
|
| 43 |
+
max_grad_norm: 1.0
|
| 44 |
+
optimizer_type: prodigy
|
| 45 |
+
scheduler_params: {}
|
| 46 |
+
scheduler_type: cosine
|
| 47 |
+
steps: 5000
|
| 48 |
+
output_dir: /train/LTX-2/packages/ltx-trainer/outputs/ltx2_v2v_ic_lora_head_swap_128_128_normal
|
| 49 |
+
seed: 42
|
| 50 |
+
training_strategy:
|
| 51 |
+
audio_latents_dir: audio_latents
|
| 52 |
+
first_frame_conditioning_p: 1.0
|
| 53 |
+
name: text_to_video
|
| 54 |
+
with_audio: false
|
| 55 |
+
validation:
|
| 56 |
+
frame_rate: 25.0
|
| 57 |
+
generate_audio: false
|
| 58 |
+
guidance_scale: 4.0
|
| 59 |
+
images:
|
| 60 |
+
- dataset/test/351912466498942.png
|
| 61 |
+
- dataset/test/6403624465969299.png
|
| 62 |
+
include_reference_in_output: true
|
| 63 |
+
inference_steps: 30
|
| 64 |
+
interval: 250
|
| 65 |
+
negative_prompt: worst quality, inconsistent motion, blurry, jittery, distorted
|
| 66 |
+
prompts:
|
| 67 |
+
- High-fidelity face swap, strictly maintaining the identity and facial features
|
| 68 |
+
from the first frame throughout the entire video.
|
| 69 |
+
- High-fidelity face swap, strictly maintaining the identity and facial features
|
| 70 |
+
from the first frame throughout the entire video.
|
| 71 |
+
reference_downscale_factor: 1
|
| 72 |
+
reference_videos: null
|
| 73 |
+
seed: 42
|
| 74 |
+
skip_initial_validation: true
|
| 75 |
+
stg_blocks:
|
| 76 |
+
- 29
|
| 77 |
+
stg_mode: stg_v
|
| 78 |
+
stg_scale: 0.0
|
| 79 |
+
video_dims: !!python/tuple
|
| 80 |
+
- 576
|
| 81 |
+
- 1024
|
| 82 |
+
- 49
|
| 83 |
+
videos_per_prompt: 1
|
| 84 |
+
wandb:
|
| 85 |
+
enabled: true
|
| 86 |
+
entity: null
|
| 87 |
+
log_validation_videos: true
|
| 88 |
+
project: ltx-2-trainer-01-02-2026-normal
|
| 89 |
+
tags:
|
| 90 |
+
- ltx2
|
| 91 |
+
- ic-lora
|
| 92 |
+
- video-to-video
|