|
{ |
|
"architectures": [ |
|
"HARTAutoEncoderWithDisc" |
|
], |
|
"beta": 0.25, |
|
"ch": 160, |
|
"ch_mult": [ |
|
1, |
|
1, |
|
2, |
|
2, |
|
4 |
|
], |
|
"default_qresi_counts": 0, |
|
"disable_quant_resi": false, |
|
"double_decoder": false, |
|
"dropout": 0.0, |
|
"freeze_codebook_for_hybrid": true, |
|
"hybrid": false, |
|
"levels": [ |
|
8, |
|
8, |
|
8, |
|
6, |
|
5 |
|
], |
|
"model_type": "hart_autoencoder_with_disc", |
|
"quant_conv_ks": 3, |
|
"quant_resi": 0.5, |
|
"quantizer_type": "var_hybrid", |
|
"share_quant_resi": 4, |
|
"test_mode": false, |
|
"torch_dtype": "float32", |
|
"transformers_version": "4.36.2", |
|
"using_znorm": false, |
|
"v_patch_nums": [ |
|
1, |
|
2, |
|
3, |
|
4, |
|
5, |
|
7, |
|
9, |
|
12, |
|
16, |
|
21, |
|
27, |
|
36, |
|
48, |
|
64 |
|
], |
|
"vocab_size": 4096, |
|
"z_channels": 32 |
|
} |
|
|