Function unet

scripts/convert_models_diffuser_to_diffusers.py:15–56

(hor)

def unet(hor):
    if hor == 128:
        down_block_types = ("DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D")
        block_out_channels = (32, 128, 256)
        up_block_types = ("UpResnetBlock1D", "UpResnetBlock1D")

    elif hor == 32:
        down_block_types = ("DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D", "DownResnetBlock1D")
        block_out_channels = (32, 64, 128, 256)
        up_block_types = ("UpResnetBlock1D", "UpResnetBlock1D", "UpResnetBlock1D")
    model = torch.load(f"/Users/bglickenhaus/Documents/diffuser/temporal_unet-hopper-mediumv2-hor{hor}.torch")
    state_dict = model.state_dict()
    config = {
        "down_block_types": down_block_types,
        "block_out_channels": block_out_channels,
        "up_block_types": up_block_types,
        "layers_per_block": 1,
        "use_timestep_embedding": True,
        "out_block_type": "OutConv1DBlock",
        "norm_num_groups": 8,
        "downsample_each_block": False,
        "in_channels": 14,
        "out_channels": 14,
        "extra_in_channels": 0,
        "time_embedding_type": "positional",
        "flip_sin_to_cos": False,
        "freq_shift": 1,
        "sample_size": 65536,
        "mid_block_type": "MidResTemporalBlock1D",
        "act_fn": "mish",
    }
    hf_value_function = UNet1DModel(**config)
    print(f"length of state dict: {len(state_dict.keys())}")
    print(f"length of value function dict: {len(hf_value_function.state_dict().keys())}")
    mapping = dict(zip(model.state_dict().keys(), hf_value_function.state_dict().keys()))
    for k, v in mapping.items():
        state_dict[v] = state_dict.pop(k)
    hf_value_function.load_state_dict(state_dict)

    torch.save(hf_value_function.state_dict(), f"hub/hopper-medium-v2/unet/hor{hor}/diffusion_pytorch_model.bin")
    with open(f"hub/hopper-medium-v2/unet/hor{hor}/config.json", "w") as f:
        json.dump(config, f)


def value_function():