trainer_2/config.json

{
    "trainer": {
        "trainer_class": "MatryoshkaBatchTopKTrainer",
        "dict_class": "MatryoshkaBatchTopKSAE",
        "lr": 5e-05,
        "steps": 4394,
        "auxk_alpha": 0.03125,
        "warmup_steps": 1000,
        "decay_start": 3515,
        "threshold_beta": 0.999,
        "threshold_start_step": 1000,
        "top_k_aux": 64,
        "seed": 0,
        "activation_dim": 128,
        "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": "unknown",
        "lm_name": "Lightricks/LTX-Video-0.9.5",
        "wandb_name": "MatryoshkaBatchTopKTrainer-Lightricks/LTX-Video-0.9.5-resid_post_layer_unknown_trainer_2",
        "submodule_name": "resid_post_layer_unknown"
    },
    "buffer": {
        "activation_dir": "/mnt/nw/home/m.yu/repos/dictionary_learning_demo/ltx_activations_vae",
        "d_submodule": 128,
        "out_batch_size": 2048,
        "num_shards": 70,
        "metadata": {
            "model_name": "Lightricks/LTX-Video-0.9.5",
            "hook_target": "vae_latent_mean",
            "d_model": 128,
            "num_frames": 321,
            "height": 256,
            "width": 256,
            "max_norm_multiple": 10,
            "total_tokens": 9063296,
            "num_shards": 70,
            "num_videos": 3454,
            "save_dtype": "float32"
        }
    }
}