vqgan-16x16 / config.json
BorelTHU's picture
Push model using huggingface_hub.
9c067ce verified
{
"ckpt_path": null,
"colorize_nlabels": null,
"decoder": {
"attn_resolutions": [
16
],
"ch": 128,
"ch_mult": [
1,
1,
2,
2,
4
],
"double_z": false,
"dropout": 0.0,
"in_channels": 3,
"num_res_blocks": 2,
"out_ch": 3,
"resolution": 256,
"z_channels": 256
},
"encoder": {
"attn_resolutions": [
16
],
"ch": 128,
"ch_mult": [
1,
1,
2,
2,
4
],
"double_z": false,
"dropout": 0.0,
"in_channels": 3,
"num_res_blocks": 2,
"out_ch": 3,
"resolution": 256,
"z_channels": 256
},
"ignore_keys": [],
"image_key": "image",
"loss": {
"disc_in_channels": 3,
"disc_num_layers": 2,
"disc_start": 150000,
"loss_d_weight": 1.0,
"loss_g_weight": 1.0,
"loss_l1_weight": 1.0,
"loss_l2_weight": 1.0,
"loss_p_weight": 3.0,
"loss_q_weight": 1.0
},
"monitor": null,
"quantize": {
"beta": 0.25,
"e_dim": 256,
"fix_codes": false,
"n_e": 16384,
"num_head": 1,
"start_quantize_steps": null,
"use_proj": false
},
"quantize_type": "basevq",
"use_connector": true
}