{ "trainer": { "trainer_class": "MatryoshkaBatchTopKTrainer", "dict_class": "MatryoshkaBatchTopKSAE", "lr": 5e-05, "steps": 244140, "auxk_alpha": 0.03125, "warmup_steps": 1000, "decay_start": 195312, "threshold_beta": 0.999, "threshold_start_step": 1000, "top_k_aux": 256, "seed": 0, "activation_dim": 512, "dict_size": 8192, "group_fractions": [ 0.03125, 0.0625, 0.125, 0.25, 0.53125 ], "group_weights": [ 0.2, 0.2, 0.2, 0.2, 0.2 ], "group_sizes": [ 256, 512, 1024, 2048, 4352 ], "k": 10, "device": "cuda:0", "layer": "encoder.mid_block", "lm_name": "Lightricks/LTX-Video", "wandb_name": "MatryoshkaBatchTopKTrainer-Lightricks/LTX-Video-resid_post_layer_encoder.mid_block_trainer_2", "submodule_name": "resid_post_layer_encoder.mid_block" }, "buffer": { "activation_dir": "./ltx_activations/encoder_mid_block", "d_submodule": 512, "out_batch_size": 2048, "num_shards": 70, "metadata": { "model_name": "Lightricks/LTX-Video", "hook_target": "vae_encoder", "hook_module": "encoder.mid_block", "d_model": 512, "num_frames": 25, "height": 256, "width": 256, "text_prompt": "", "shift": 1.0, "max_norm_multiple": 10, "total_tokens": 884224, "num_shards": 70, "num_videos": 3454, "save_dtype": "float32" } } }