LTX-2.5 LoRA β Tessy
LoRA de identidad entrenado sobre LTX-2.5 con 60 imagenes de "tessy".
Detalles del entrenamiento
- Base model: Lightricks/LTX-2.5 (dev-transformer-bf16)
- Training mode: LoRA
- Rank / Alpha: 32 / 32
- Steps: 1500
- Learning rate: 1e-4 (linear)
- Dataset: 60 imagenes, resolution bucket 960x544x1
- Modalidad: imagenes (1 frame por sample)
- Trigger word: tessy
Uso
tessy, close-up portrait of a woman with long dark hair, gentle expression, natural lighting
Archivos
- checkpoints/lora_weights_step_01500.safetensors β LoRA final
- checkpoints/lora_weights_step_*.safetensors β Checkpoints intermedios
- samples/ β Videos de validacion
- training_config.yaml β Config completo
Model tree for aztro/ltx-2.5-tessy-lora
Base model
Lightricks/LTX-2.5