{ "architecture": "voxcpm2", "audio_vae_config": { "decoder_dim": 2048, "decoder_rates": [ 8, 6, 5, 2, 2, 2 ], "encoder_dim": 128, "encoder_rates": [ 2, 5, 8, 8 ], "latent_dim": 64, "out_sample_rate": 48000, "sample_rate": 16000, "sr_bin_boundaries": [ 20000, 30000, 40000 ] }, "device": "cuda", "dit_config": { "cfm_config": { "inference_cfg_rate": 2.0, "sigma_min": 1e-06, "solver": "euler", "t_scheduler": "log-norm" }, "ffn_dim": 4096, "hidden_dim": 1024, "kv_channels": 128, "mean_mode": false, "num_heads": 16, "num_layers": 12 }, "dtype": "bfloat16", "encoder_config": { "ffn_dim": 4096, "hidden_dim": 1024, "kv_channels": 128, "num_heads": 16, "num_layers": 12 }, "feat_dim": 64, "lm_config": { "bos_token_id": 1, "dim_model_base": 256, "eos_token_id": 2, "hidden_size": 2048, "intermediate_size": 6144, "kv_channels": 128, "max_position_embeddings": 32768, "num_attention_heads": 16, "num_hidden_layers": 28, "num_key_value_heads": 2, "rms_norm_eps": 1e-05, "rope_scaling": { "long_factor": [ 0.9977997200264581, 1.014658295992452, 1.0349680404997148, 1.059429246056193, 1.0888815016813513, 1.1243301355211495, 1.166977103606075, 1.2182568066927284, 1.2798772354275727, 1.3538666751582975, 1.4426259039919596, 1.5489853358570191, 1.6762658237220625, 1.8283407612492941, 2.0096956085876183, 2.225478927469756, 2.481536379650452, 2.784415934557119, 3.1413289096347365, 3.560047844772632, 4.048719380066383, 4.615569542115128, 5.2684819496549835, 6.014438591970396, 6.858830049237097, 7.804668263503327, 8.851768731513417, 9.99600492938444, 11.228766118181639, 12.536757560834843, 13.902257701387796, 15.303885189125953, 16.717837610115794, 18.119465097853947, 19.484965238406907, 20.792956681060105, 22.02571786985731, 23.16995406772833, 24.217054535738416, 25.16289275000465, 26.007284207271347, 26.753240849586767, 27.40615325712662, 27.973003419175363, 28.461674954469114, 28.880393889607006, 29.237306864684626, 29.540186419591297, 29.79624387177199, 30.01202719065413, 30.193382037992453, 30.34545697551969, 30.47273746338473, 30.579096895249787, 30.66785612408345, 30.741845563814174, 30.80346599254902, 30.85474569563567, 30.897392663720595, 30.932841297560394, 30.962293553185553, 30.986754758742034, 31.007064503249293, 31.02392307921529 ], "original_max_position_embeddings": 32768, "short_factor": [ 0.9977997200264581, 1.014658295992452, 1.0349680404997148, 1.059429246056193, 1.0888815016813513, 1.1243301355211495, 1.166977103606075, 1.2182568066927284, 1.2798772354275727, 1.3538666751582975, 1.4426259039919596, 1.5489853358570191, 1.6762658237220625, 1.8283407612492941, 2.0096956085876183, 2.225478927469756, 2.481536379650452, 2.784415934557119, 3.1413289096347365, 3.560047844772632, 4.048719380066383, 4.615569542115128, 5.2684819496549835, 6.014438591970396, 6.858830049237097, 7.804668263503327, 8.851768731513417, 9.99600492938444, 11.228766118181639, 12.536757560834843, 13.902257701387796, 15.303885189125953, 16.717837610115794, 18.119465097853947, 19.484965238406907, 20.792956681060105, 22.02571786985731, 23.16995406772833, 24.217054535738416, 25.16289275000465, 26.007284207271347, 26.753240849586767, 27.40615325712662, 27.973003419175363, 28.461674954469114, 28.880393889607006, 29.237306864684626, 29.540186419591297, 29.79624387177199, 30.01202719065413, 30.193382037992453, 30.34545697551969, 30.47273746338473, 30.579096895249787, 30.66785612408345, 30.741845563814174, 30.80346599254902, 30.85474569563567, 30.897392663720595, 30.932841297560394, 30.962293553185553, 30.986754758742034, 31.007064503249293, 31.02392307921529 ], "type": "longrope" }, "rope_theta": 10000, "scale_depth": 1.4, "scale_emb": 12, "use_mup": false, "vocab_size": 73448 }, "max_length": 8192, "patch_size": 4, "quantization_config": { "config_groups": { "group_0": { "format": "nvfp4-pack-quantized", "input_activations": null, "output_activations": null, "targets": [ "Linear" ], "weights": { "actorder": null, "block_structure": null, "dynamic": false, "group_size": 16, "num_bits": 4, "observer": "memoryless_minmax", "observer_kwargs": {}, "scale_dtype": "torch.float8_e4m3fn", "strategy": "tensor_group", "symmetric": true, "type": "float", "zp_dtype": null } } }, "format": "nvfp4-pack-quantized", "global_compression_ratio": null, "ignore": [ "base_lm.layers.0.self_attn.v_proj", "base_lm.layers.0.mlp.down_proj", "base_lm.layers.1.self_attn.v_proj", "base_lm.layers.1.mlp.down_proj", "base_lm.layers.2.self_attn.v_proj", "base_lm.layers.2.mlp.down_proj", "base_lm.layers.3.self_attn.v_proj", "base_lm.layers.3.mlp.down_proj", "base_lm.layers.4.self_attn.v_proj", "base_lm.layers.4.mlp.down_proj", "base_lm.layers.5.self_attn.v_proj", "base_lm.layers.5.mlp.down_proj", "base_lm.layers.6.self_attn.v_proj", "base_lm.layers.6.mlp.down_proj", "base_lm.layers.7.self_attn.v_proj", "base_lm.layers.7.mlp.down_proj", "base_lm.layers.8.self_attn.v_proj", "base_lm.layers.8.mlp.down_proj", "base_lm.layers.9.self_attn.v_proj", "base_lm.layers.9.mlp.down_proj", "base_lm.layers.10.self_attn.v_proj", "base_lm.layers.10.mlp.down_proj", "base_lm.layers.11.self_attn.v_proj", "base_lm.layers.11.mlp.down_proj", "base_lm.layers.12.self_attn.v_proj", "base_lm.layers.12.mlp.down_proj", "base_lm.layers.13.self_attn.v_proj", "base_lm.layers.13.mlp.down_proj", "base_lm.layers.14.self_attn.v_proj", "base_lm.layers.14.mlp.down_proj", "base_lm.layers.15.self_attn.v_proj", "base_lm.layers.15.mlp.down_proj", "base_lm.layers.16.self_attn.v_proj", "base_lm.layers.16.mlp.down_proj", "base_lm.layers.17.self_attn.v_proj", "base_lm.layers.17.mlp.down_proj", "base_lm.layers.18.self_attn.v_proj", "base_lm.layers.18.mlp.down_proj", "base_lm.layers.19.self_attn.v_proj", "base_lm.layers.19.mlp.down_proj", "base_lm.layers.20.self_attn.v_proj", "base_lm.layers.20.mlp.down_proj", "base_lm.layers.21.self_attn.v_proj", "base_lm.layers.21.mlp.down_proj", "base_lm.layers.22.self_attn.v_proj", "base_lm.layers.22.mlp.down_proj", "base_lm.layers.23.self_attn.v_proj", "base_lm.layers.23.mlp.down_proj", "base_lm.layers.24.self_attn.v_proj", "base_lm.layers.24.mlp.down_proj", "base_lm.layers.25.self_attn.v_proj", "base_lm.layers.25.mlp.down_proj", "base_lm.layers.26.self_attn.v_proj", "base_lm.layers.26.mlp.down_proj", "base_lm.layers.27.self_attn.v_proj", "base_lm.layers.27.mlp.down_proj", "residual_lm.layers.0.self_attn.v_proj", "residual_lm.layers.0.mlp.down_proj", "residual_lm.layers.1.self_attn.v_proj", "residual_lm.layers.1.mlp.down_proj", "residual_lm.layers.2.self_attn.v_proj", "residual_lm.layers.2.mlp.down_proj", "residual_lm.layers.3.self_attn.v_proj", "residual_lm.layers.3.mlp.down_proj", "residual_lm.layers.4.self_attn.v_proj", "residual_lm.layers.4.mlp.down_proj", "residual_lm.layers.5.self_attn.v_proj", "residual_lm.layers.5.mlp.down_proj", "residual_lm.layers.6.self_attn.v_proj", "residual_lm.layers.6.mlp.down_proj", "residual_lm.layers.7.self_attn.v_proj", "residual_lm.layers.7.mlp.down_proj", "feat_encoder.in_proj", "feat_encoder.encoder.layers.0.self_attn.q_proj", "feat_encoder.encoder.layers.0.self_attn.k_proj", "feat_encoder.encoder.layers.0.self_attn.v_proj", "feat_encoder.encoder.layers.0.self_attn.o_proj", "feat_encoder.encoder.layers.0.mlp.gate_proj", "feat_encoder.encoder.layers.0.mlp.up_proj", "feat_encoder.encoder.layers.0.mlp.down_proj", "feat_encoder.encoder.layers.1.self_attn.q_proj", "feat_encoder.encoder.layers.1.self_attn.k_proj", "feat_encoder.encoder.layers.1.self_attn.v_proj", "feat_encoder.encoder.layers.1.self_attn.o_proj", "feat_encoder.encoder.layers.1.mlp.gate_proj", "feat_encoder.encoder.layers.1.mlp.up_proj", "feat_encoder.encoder.layers.1.mlp.down_proj", "feat_encoder.encoder.layers.2.self_attn.q_proj", "feat_encoder.encoder.layers.2.self_attn.k_proj", "feat_encoder.encoder.layers.2.self_attn.v_proj", "feat_encoder.encoder.layers.2.self_attn.o_proj", "feat_encoder.encoder.layers.2.mlp.gate_proj", "feat_encoder.encoder.layers.2.mlp.up_proj", "feat_encoder.encoder.layers.2.mlp.down_proj", "feat_encoder.encoder.layers.3.self_attn.q_proj", "feat_encoder.encoder.layers.3.self_attn.k_proj", "feat_encoder.encoder.layers.3.self_attn.v_proj", "feat_encoder.encoder.layers.3.self_attn.o_proj", "feat_encoder.encoder.layers.3.mlp.gate_proj", "feat_encoder.encoder.layers.3.mlp.up_proj", "feat_encoder.encoder.layers.3.mlp.down_proj", "feat_encoder.encoder.layers.4.self_attn.q_proj", "feat_encoder.encoder.layers.4.self_attn.k_proj", "feat_encoder.encoder.layers.4.self_attn.v_proj", "feat_encoder.encoder.layers.4.self_attn.o_proj", "feat_encoder.encoder.layers.4.mlp.gate_proj", "feat_encoder.encoder.layers.4.mlp.up_proj", "feat_encoder.encoder.layers.4.mlp.down_proj", "feat_encoder.encoder.layers.5.self_attn.q_proj", "feat_encoder.encoder.layers.5.self_attn.k_proj", "feat_encoder.encoder.layers.5.self_attn.v_proj", "feat_encoder.encoder.layers.5.self_attn.o_proj", "feat_encoder.encoder.layers.5.mlp.gate_proj", "feat_encoder.encoder.layers.5.mlp.up_proj", "feat_encoder.encoder.layers.5.mlp.down_proj", "feat_encoder.encoder.layers.6.self_attn.q_proj", "feat_encoder.encoder.layers.6.self_attn.k_proj", "feat_encoder.encoder.layers.6.self_attn.v_proj", "feat_encoder.encoder.layers.6.self_attn.o_proj", "feat_encoder.encoder.layers.6.mlp.gate_proj", "feat_encoder.encoder.layers.6.mlp.up_proj", "feat_encoder.encoder.layers.6.mlp.down_proj", "feat_encoder.encoder.layers.7.self_attn.q_proj", "feat_encoder.encoder.layers.7.self_attn.k_proj", "feat_encoder.encoder.layers.7.self_attn.v_proj", "feat_encoder.encoder.layers.7.self_attn.o_proj", "feat_encoder.encoder.layers.7.mlp.gate_proj", "feat_encoder.encoder.layers.7.mlp.up_proj", "feat_encoder.encoder.layers.7.mlp.down_proj", "feat_encoder.encoder.layers.8.self_attn.q_proj", "feat_encoder.encoder.layers.8.self_attn.k_proj", "feat_encoder.encoder.layers.8.self_attn.v_proj", "feat_encoder.encoder.layers.8.self_attn.o_proj", "feat_encoder.encoder.layers.8.mlp.gate_proj", "feat_encoder.encoder.layers.8.mlp.up_proj", "feat_encoder.encoder.layers.8.mlp.down_proj", "feat_encoder.encoder.layers.9.self_attn.q_proj", "feat_encoder.encoder.layers.9.self_attn.k_proj", "feat_encoder.encoder.layers.9.self_attn.v_proj", "feat_encoder.encoder.layers.9.self_attn.o_proj", "feat_encoder.encoder.layers.9.mlp.gate_proj", "feat_encoder.encoder.layers.9.mlp.up_proj", "feat_encoder.encoder.layers.9.mlp.down_proj", "feat_encoder.encoder.layers.10.self_attn.q_proj", "feat_encoder.encoder.layers.10.self_attn.k_proj", "feat_encoder.encoder.layers.10.self_attn.v_proj", "feat_encoder.encoder.layers.10.self_attn.o_proj", "feat_encoder.encoder.layers.10.mlp.gate_proj", "feat_encoder.encoder.layers.10.mlp.up_proj", "feat_encoder.encoder.layers.10.mlp.down_proj", "feat_encoder.encoder.layers.11.self_attn.q_proj", "feat_encoder.encoder.layers.11.self_attn.k_proj", "feat_encoder.encoder.layers.11.self_attn.v_proj", "feat_encoder.encoder.layers.11.self_attn.o_proj", "feat_encoder.encoder.layers.11.mlp.gate_proj", "feat_encoder.encoder.layers.11.mlp.up_proj", "feat_encoder.encoder.layers.11.mlp.down_proj", "feat_decoder.estimator.in_proj", "feat_decoder.estimator.cond_proj", "feat_decoder.estimator.out_proj", "feat_decoder.estimator.time_mlp.linear_1", "feat_decoder.estimator.time_mlp.linear_2", "feat_decoder.estimator.delta_time_mlp.linear_1", "feat_decoder.estimator.delta_time_mlp.linear_2", "feat_decoder.estimator.decoder.layers.0.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.0.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.0.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.0.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.0.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.0.mlp.up_proj", "feat_decoder.estimator.decoder.layers.0.mlp.down_proj", "feat_decoder.estimator.decoder.layers.1.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.1.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.1.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.1.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.1.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.1.mlp.up_proj", "feat_decoder.estimator.decoder.layers.1.mlp.down_proj", "feat_decoder.estimator.decoder.layers.2.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.2.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.2.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.2.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.2.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.2.mlp.up_proj", "feat_decoder.estimator.decoder.layers.2.mlp.down_proj", "feat_decoder.estimator.decoder.layers.3.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.3.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.3.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.3.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.3.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.3.mlp.up_proj", "feat_decoder.estimator.decoder.layers.3.mlp.down_proj", "feat_decoder.estimator.decoder.layers.4.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.4.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.4.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.4.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.4.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.4.mlp.up_proj", "feat_decoder.estimator.decoder.layers.4.mlp.down_proj", "feat_decoder.estimator.decoder.layers.5.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.5.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.5.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.5.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.5.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.5.mlp.up_proj", "feat_decoder.estimator.decoder.layers.5.mlp.down_proj", "feat_decoder.estimator.decoder.layers.6.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.6.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.6.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.6.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.6.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.6.mlp.up_proj", "feat_decoder.estimator.decoder.layers.6.mlp.down_proj", "feat_decoder.estimator.decoder.layers.7.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.7.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.7.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.7.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.7.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.7.mlp.up_proj", "feat_decoder.estimator.decoder.layers.7.mlp.down_proj", "feat_decoder.estimator.decoder.layers.8.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.8.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.8.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.8.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.8.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.8.mlp.up_proj", "feat_decoder.estimator.decoder.layers.8.mlp.down_proj", "feat_decoder.estimator.decoder.layers.9.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.9.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.9.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.9.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.9.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.9.mlp.up_proj", "feat_decoder.estimator.decoder.layers.9.mlp.down_proj", "feat_decoder.estimator.decoder.layers.10.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.10.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.10.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.10.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.10.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.10.mlp.up_proj", "feat_decoder.estimator.decoder.layers.10.mlp.down_proj", "feat_decoder.estimator.decoder.layers.11.self_attn.q_proj", "feat_decoder.estimator.decoder.layers.11.self_attn.k_proj", "feat_decoder.estimator.decoder.layers.11.self_attn.v_proj", "feat_decoder.estimator.decoder.layers.11.self_attn.o_proj", "feat_decoder.estimator.decoder.layers.11.mlp.gate_proj", "feat_decoder.estimator.decoder.layers.11.mlp.up_proj", "feat_decoder.estimator.decoder.layers.11.mlp.down_proj", "fsq_layer.in_proj", "fsq_layer.out_proj", "enc_to_lm_proj", "lm_to_dit_proj", "res_to_dit_proj", "fusion_concat_proj", "stop_proj", "stop_head" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed", "sparsity_config": {}, "transform_config": {}, "version": "0.14.0.1" }, "residual_lm_no_rope": true, "residual_lm_num_layers": 8, "scalar_quantization_latent_dim": 512, "scalar_quantization_scale": 9 }