| { | |
| "data": { | |
| "text_length": 1024, | |
| "audio_length": 3072, | |
| "channels": 9, | |
| "text_pad_value": 0, | |
| "audio_eos_value": 1024, | |
| "audio_pad_value": 1025, | |
| "audio_bos_value": 1026, | |
| "delay_pattern": [ | |
| 0, | |
| 8, | |
| 9, | |
| 10, | |
| 11, | |
| 12, | |
| 13, | |
| 14, | |
| 15 | |
| ] | |
| }, | |
| "model": { | |
| "encoder": { | |
| "n_layer": 12, | |
| "n_embd": 1024, | |
| "n_hidden": 4096, | |
| "n_head": 16, | |
| "head_dim": 128 | |
| }, | |
| "decoder": { | |
| "n_layer": 18, | |
| "n_embd": 2048, | |
| "n_hidden": 8192, | |
| "gqa_query_heads": 16, | |
| "cross_query_heads": 16, | |
| "kv_heads": 4, | |
| "gqa_head_dim": 128, | |
| "cross_head_dim": 128 | |
| }, | |
| "src_vocab_size": 256, | |
| "tgt_vocab_size": 1028, | |
| "dropout": 0.0 | |
| }, | |
| "quantization": { | |
| "group_size": 64, | |
| "bits": 3, | |
| "model.encoder.embedding": false, | |
| "model.encoder.dropout": false, | |
| "model.encoder.layers.0.pre_sa_norm": false, | |
| "model.encoder.layers.0.self_attention.q_proj": false, | |
| "model.encoder.layers.0.self_attention.k_proj": false, | |
| "model.encoder.layers.0.self_attention.v_proj": false, | |
| "model.encoder.layers.0.self_attention.o_proj": false, | |
| "model.encoder.layers.0.self_attention.rotary_emb": false, | |
| "model.encoder.layers.0.post_sa_norm": false, | |
| "model.encoder.layers.0.mlp.wi_fused": false, | |
| "model.encoder.layers.0.mlp.activation_fn_0": false, | |
| "model.encoder.layers.0.mlp.activation_fn_1": false, | |
| "model.encoder.layers.0.mlp.dropout": false, | |
| "model.encoder.layers.0.mlp.wo": false, | |
| "model.encoder.layers.0.dropout": false, | |
| "model.encoder.layers.1.pre_sa_norm": false, | |
| "model.encoder.layers.1.self_attention.q_proj": false, | |
| "model.encoder.layers.1.self_attention.k_proj": false, | |
| "model.encoder.layers.1.self_attention.v_proj": false, | |
| "model.encoder.layers.1.self_attention.o_proj": false, | |
| "model.encoder.layers.1.self_attention.rotary_emb": false, | |
| "model.encoder.layers.1.post_sa_norm": false, | |
| "model.encoder.layers.1.mlp.wi_fused": false, | |
| "model.encoder.layers.1.mlp.activation_fn_0": false, | |
| "model.encoder.layers.1.mlp.activation_fn_1": false, | |
| "model.encoder.layers.1.mlp.dropout": false, | |
| "model.encoder.layers.1.mlp.wo": false, | |
| "model.encoder.layers.1.dropout": false, | |
| "model.encoder.layers.2.pre_sa_norm": false, | |
| "model.encoder.layers.2.self_attention.q_proj": false, | |
| "model.encoder.layers.2.self_attention.k_proj": false, | |
| "model.encoder.layers.2.self_attention.v_proj": false, | |
| "model.encoder.layers.2.self_attention.o_proj": false, | |
| "model.encoder.layers.2.self_attention.rotary_emb": false, | |
| "model.encoder.layers.2.post_sa_norm": false, | |
| "model.encoder.layers.2.mlp.wi_fused": false, | |
| "model.encoder.layers.2.mlp.activation_fn_0": false, | |
| "model.encoder.layers.2.mlp.activation_fn_1": false, | |
| "model.encoder.layers.2.mlp.dropout": false, | |
| "model.encoder.layers.2.mlp.wo": false, | |
| "model.encoder.layers.2.dropout": false, | |
| "model.encoder.layers.3.pre_sa_norm": false, | |
| "model.encoder.layers.3.self_attention.q_proj": false, | |
| "model.encoder.layers.3.self_attention.k_proj": false, | |
| "model.encoder.layers.3.self_attention.v_proj": false, | |
| "model.encoder.layers.3.self_attention.o_proj": false, | |
| "model.encoder.layers.3.self_attention.rotary_emb": false, | |
| "model.encoder.layers.3.post_sa_norm": false, | |
| "model.encoder.layers.3.mlp.wi_fused": false, | |
| "model.encoder.layers.3.mlp.activation_fn_0": false, | |
| "model.encoder.layers.3.mlp.activation_fn_1": false, | |
| "model.encoder.layers.3.mlp.dropout": false, | |
| "model.encoder.layers.3.mlp.wo": false, | |
| "model.encoder.layers.3.dropout": false, | |
| "model.encoder.layers.4.pre_sa_norm": false, | |
| "model.encoder.layers.4.self_attention.q_proj": false, | |
| "model.encoder.layers.4.self_attention.k_proj": false, | |
| "model.encoder.layers.4.self_attention.v_proj": false, | |
| "model.encoder.layers.4.self_attention.o_proj": false, | |
| "model.encoder.layers.4.self_attention.rotary_emb": false, | |
| "model.encoder.layers.4.post_sa_norm": false, | |
| "model.encoder.layers.4.mlp.wi_fused": false, | |
| "model.encoder.layers.4.mlp.activation_fn_0": false, | |
| "model.encoder.layers.4.mlp.activation_fn_1": false, | |
| "model.encoder.layers.4.mlp.dropout": false, | |
| "model.encoder.layers.4.mlp.wo": false, | |
| "model.encoder.layers.4.dropout": false, | |
| "model.encoder.layers.5.pre_sa_norm": false, | |
| "model.encoder.layers.5.self_attention.q_proj": false, | |
| "model.encoder.layers.5.self_attention.k_proj": false, | |
| "model.encoder.layers.5.self_attention.v_proj": false, | |
| "model.encoder.layers.5.self_attention.o_proj": false, | |
| "model.encoder.layers.5.self_attention.rotary_emb": false, | |
| "model.encoder.layers.5.post_sa_norm": false, | |
| "model.encoder.layers.5.mlp.wi_fused": false, | |
| "model.encoder.layers.5.mlp.activation_fn_0": false, | |
| "model.encoder.layers.5.mlp.activation_fn_1": false, | |
| "model.encoder.layers.5.mlp.dropout": false, | |
| "model.encoder.layers.5.mlp.wo": false, | |
| "model.encoder.layers.5.dropout": false, | |
| "model.encoder.layers.6.pre_sa_norm": false, | |
| "model.encoder.layers.6.self_attention.q_proj": false, | |
| "model.encoder.layers.6.self_attention.k_proj": false, | |
| "model.encoder.layers.6.self_attention.v_proj": false, | |
| "model.encoder.layers.6.self_attention.o_proj": false, | |
| "model.encoder.layers.6.self_attention.rotary_emb": false, | |
| "model.encoder.layers.6.post_sa_norm": false, | |
| "model.encoder.layers.6.mlp.wi_fused": false, | |
| "model.encoder.layers.6.mlp.activation_fn_0": false, | |
| "model.encoder.layers.6.mlp.activation_fn_1": false, | |
| "model.encoder.layers.6.mlp.dropout": false, | |
| "model.encoder.layers.6.mlp.wo": false, | |
| "model.encoder.layers.6.dropout": false, | |
| "model.encoder.layers.7.pre_sa_norm": false, | |
| "model.encoder.layers.7.self_attention.q_proj": false, | |
| "model.encoder.layers.7.self_attention.k_proj": false, | |
| "model.encoder.layers.7.self_attention.v_proj": false, | |
| "model.encoder.layers.7.self_attention.o_proj": false, | |
| "model.encoder.layers.7.self_attention.rotary_emb": false, | |
| "model.encoder.layers.7.post_sa_norm": false, | |
| "model.encoder.layers.7.mlp.wi_fused": false, | |
| "model.encoder.layers.7.mlp.activation_fn_0": false, | |
| "model.encoder.layers.7.mlp.activation_fn_1": false, | |
| "model.encoder.layers.7.mlp.dropout": false, | |
| "model.encoder.layers.7.mlp.wo": false, | |
| "model.encoder.layers.7.dropout": false, | |
| "model.encoder.layers.8.pre_sa_norm": false, | |
| "model.encoder.layers.8.self_attention.q_proj": false, | |
| "model.encoder.layers.8.self_attention.k_proj": false, | |
| "model.encoder.layers.8.self_attention.v_proj": false, | |
| "model.encoder.layers.8.self_attention.o_proj": false, | |
| "model.encoder.layers.8.self_attention.rotary_emb": false, | |
| "model.encoder.layers.8.post_sa_norm": false, | |
| "model.encoder.layers.8.mlp.wi_fused": false, | |
| "model.encoder.layers.8.mlp.activation_fn_0": false, | |
| "model.encoder.layers.8.mlp.activation_fn_1": false, | |
| "model.encoder.layers.8.mlp.dropout": false, | |
| "model.encoder.layers.8.mlp.wo": false, | |
| "model.encoder.layers.8.dropout": false, | |
| "model.encoder.layers.9.pre_sa_norm": false, | |
| "model.encoder.layers.9.self_attention.q_proj": false, | |
| "model.encoder.layers.9.self_attention.k_proj": false, | |
| "model.encoder.layers.9.self_attention.v_proj": false, | |
| "model.encoder.layers.9.self_attention.o_proj": false, | |
| "model.encoder.layers.9.self_attention.rotary_emb": false, | |
| "model.encoder.layers.9.post_sa_norm": false, | |
| "model.encoder.layers.9.mlp.wi_fused": false, | |
| "model.encoder.layers.9.mlp.activation_fn_0": false, | |
| "model.encoder.layers.9.mlp.activation_fn_1": false, | |
| "model.encoder.layers.9.mlp.dropout": false, | |
| "model.encoder.layers.9.mlp.wo": false, | |
| "model.encoder.layers.9.dropout": false, | |
| "model.encoder.layers.10.pre_sa_norm": false, | |
| "model.encoder.layers.10.self_attention.q_proj": false, | |
| "model.encoder.layers.10.self_attention.k_proj": false, | |
| "model.encoder.layers.10.self_attention.v_proj": false, | |
| "model.encoder.layers.10.self_attention.o_proj": false, | |
| "model.encoder.layers.10.self_attention.rotary_emb": false, | |
| "model.encoder.layers.10.post_sa_norm": false, | |
| "model.encoder.layers.10.mlp.wi_fused": false, | |
| "model.encoder.layers.10.mlp.activation_fn_0": false, | |
| "model.encoder.layers.10.mlp.activation_fn_1": false, | |
| "model.encoder.layers.10.mlp.dropout": false, | |
| "model.encoder.layers.10.mlp.wo": false, | |
| "model.encoder.layers.10.dropout": false, | |
| "model.encoder.layers.11.pre_sa_norm": false, | |
| "model.encoder.layers.11.self_attention.q_proj": false, | |
| "model.encoder.layers.11.self_attention.k_proj": false, | |
| "model.encoder.layers.11.self_attention.v_proj": false, | |
| "model.encoder.layers.11.self_attention.o_proj": false, | |
| "model.encoder.layers.11.self_attention.rotary_emb": false, | |
| "model.encoder.layers.11.post_sa_norm": false, | |
| "model.encoder.layers.11.mlp.wi_fused": false, | |
| "model.encoder.layers.11.mlp.activation_fn_0": false, | |
| "model.encoder.layers.11.mlp.activation_fn_1": false, | |
| "model.encoder.layers.11.mlp.dropout": false, | |
| "model.encoder.layers.11.mlp.wo": false, | |
| "model.encoder.layers.11.dropout": false, | |
| "model.encoder.norm": false, | |
| "model.decoder.embeddings.0": false, | |
| "model.decoder.embeddings.1": false, | |
| "model.decoder.embeddings.2": false, | |
| "model.decoder.embeddings.3": false, | |
| "model.decoder.embeddings.4": false, | |
| "model.decoder.embeddings.5": false, | |
| "model.decoder.embeddings.6": false, | |
| "model.decoder.embeddings.7": false, | |
| "model.decoder.embeddings.8": false, | |
| "model.decoder.dropout": false, | |
| "model.decoder.layers.0.pre_sa_norm": false, | |
| "model.decoder.layers.0.pre_ca_norm": false, | |
| "model.decoder.layers.0.pre_mlp_norm": false, | |
| "model.decoder.layers.0.self_attention.q_proj": false, | |
| "model.decoder.layers.0.self_attention.k_proj": false, | |
| "model.decoder.layers.0.self_attention.v_proj": false, | |
| "model.decoder.layers.0.self_attention.o_proj": false, | |
| "model.decoder.layers.0.self_attention.rotary_emb": false, | |
| "model.decoder.layers.0.cross_attention.q_proj": false, | |
| "model.decoder.layers.0.cross_attention.k_proj": false, | |
| "model.decoder.layers.0.cross_attention.v_proj": false, | |
| "model.decoder.layers.0.cross_attention.o_proj": false, | |
| "model.decoder.layers.0.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.0.mlp.wi_fused": false, | |
| "model.decoder.layers.0.mlp.activation_fn_0": false, | |
| "model.decoder.layers.0.mlp.activation_fn_1": false, | |
| "model.decoder.layers.0.mlp.dropout": false, | |
| "model.decoder.layers.0.mlp.wo": false, | |
| "model.decoder.layers.1.pre_sa_norm": false, | |
| "model.decoder.layers.1.pre_ca_norm": false, | |
| "model.decoder.layers.1.pre_mlp_norm": false, | |
| "model.decoder.layers.1.self_attention.q_proj": false, | |
| "model.decoder.layers.1.self_attention.k_proj": false, | |
| "model.decoder.layers.1.self_attention.v_proj": false, | |
| "model.decoder.layers.1.self_attention.o_proj": false, | |
| "model.decoder.layers.1.self_attention.rotary_emb": false, | |
| "model.decoder.layers.1.cross_attention.q_proj": false, | |
| "model.decoder.layers.1.cross_attention.k_proj": false, | |
| "model.decoder.layers.1.cross_attention.v_proj": false, | |
| "model.decoder.layers.1.cross_attention.o_proj": false, | |
| "model.decoder.layers.1.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.1.mlp.wi_fused": false, | |
| "model.decoder.layers.1.mlp.activation_fn_0": false, | |
| "model.decoder.layers.1.mlp.activation_fn_1": false, | |
| "model.decoder.layers.1.mlp.dropout": false, | |
| "model.decoder.layers.1.mlp.wo": false, | |
| "model.decoder.layers.2.pre_sa_norm": false, | |
| "model.decoder.layers.2.pre_ca_norm": false, | |
| "model.decoder.layers.2.pre_mlp_norm": false, | |
| "model.decoder.layers.2.self_attention.q_proj": false, | |
| "model.decoder.layers.2.self_attention.k_proj": false, | |
| "model.decoder.layers.2.self_attention.v_proj": false, | |
| "model.decoder.layers.2.self_attention.o_proj": false, | |
| "model.decoder.layers.2.self_attention.rotary_emb": false, | |
| "model.decoder.layers.2.cross_attention.q_proj": false, | |
| "model.decoder.layers.2.cross_attention.k_proj": false, | |
| "model.decoder.layers.2.cross_attention.v_proj": false, | |
| "model.decoder.layers.2.cross_attention.o_proj": false, | |
| "model.decoder.layers.2.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.2.mlp.wi_fused": false, | |
| "model.decoder.layers.2.mlp.activation_fn_0": false, | |
| "model.decoder.layers.2.mlp.activation_fn_1": false, | |
| "model.decoder.layers.2.mlp.dropout": false, | |
| "model.decoder.layers.2.mlp.wo": false, | |
| "model.decoder.layers.3.pre_sa_norm": false, | |
| "model.decoder.layers.3.pre_ca_norm": false, | |
| "model.decoder.layers.3.pre_mlp_norm": false, | |
| "model.decoder.layers.3.self_attention.q_proj": false, | |
| "model.decoder.layers.3.self_attention.k_proj": false, | |
| "model.decoder.layers.3.self_attention.v_proj": false, | |
| "model.decoder.layers.3.self_attention.o_proj": false, | |
| "model.decoder.layers.3.self_attention.rotary_emb": false, | |
| "model.decoder.layers.3.cross_attention.q_proj": false, | |
| "model.decoder.layers.3.cross_attention.k_proj": false, | |
| "model.decoder.layers.3.cross_attention.v_proj": false, | |
| "model.decoder.layers.3.cross_attention.o_proj": false, | |
| "model.decoder.layers.3.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.3.mlp.wi_fused": false, | |
| "model.decoder.layers.3.mlp.activation_fn_0": false, | |
| "model.decoder.layers.3.mlp.activation_fn_1": false, | |
| "model.decoder.layers.3.mlp.dropout": false, | |
| "model.decoder.layers.3.mlp.wo": false, | |
| "model.decoder.layers.4.pre_sa_norm": false, | |
| "model.decoder.layers.4.pre_ca_norm": false, | |
| "model.decoder.layers.4.pre_mlp_norm": false, | |
| "model.decoder.layers.4.self_attention.q_proj": false, | |
| "model.decoder.layers.4.self_attention.k_proj": false, | |
| "model.decoder.layers.4.self_attention.v_proj": false, | |
| "model.decoder.layers.4.self_attention.o_proj": false, | |
| "model.decoder.layers.4.self_attention.rotary_emb": false, | |
| "model.decoder.layers.4.cross_attention.q_proj": false, | |
| "model.decoder.layers.4.cross_attention.k_proj": false, | |
| "model.decoder.layers.4.cross_attention.v_proj": false, | |
| "model.decoder.layers.4.cross_attention.o_proj": false, | |
| "model.decoder.layers.4.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.4.mlp.wi_fused": false, | |
| "model.decoder.layers.4.mlp.activation_fn_0": false, | |
| "model.decoder.layers.4.mlp.activation_fn_1": false, | |
| "model.decoder.layers.4.mlp.dropout": false, | |
| "model.decoder.layers.4.mlp.wo": false, | |
| "model.decoder.layers.5.pre_sa_norm": false, | |
| "model.decoder.layers.5.pre_ca_norm": false, | |
| "model.decoder.layers.5.pre_mlp_norm": false, | |
| "model.decoder.layers.5.self_attention.q_proj": false, | |
| "model.decoder.layers.5.self_attention.k_proj": false, | |
| "model.decoder.layers.5.self_attention.v_proj": false, | |
| "model.decoder.layers.5.self_attention.o_proj": false, | |
| "model.decoder.layers.5.self_attention.rotary_emb": false, | |
| "model.decoder.layers.5.cross_attention.q_proj": false, | |
| "model.decoder.layers.5.cross_attention.k_proj": false, | |
| "model.decoder.layers.5.cross_attention.v_proj": false, | |
| "model.decoder.layers.5.cross_attention.o_proj": false, | |
| "model.decoder.layers.5.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.5.mlp.wi_fused": false, | |
| "model.decoder.layers.5.mlp.activation_fn_0": false, | |
| "model.decoder.layers.5.mlp.activation_fn_1": false, | |
| "model.decoder.layers.5.mlp.dropout": false, | |
| "model.decoder.layers.5.mlp.wo": false, | |
| "model.decoder.layers.6.pre_sa_norm": false, | |
| "model.decoder.layers.6.pre_ca_norm": false, | |
| "model.decoder.layers.6.pre_mlp_norm": false, | |
| "model.decoder.layers.6.self_attention.q_proj": false, | |
| "model.decoder.layers.6.self_attention.k_proj": false, | |
| "model.decoder.layers.6.self_attention.v_proj": false, | |
| "model.decoder.layers.6.self_attention.o_proj": false, | |
| "model.decoder.layers.6.self_attention.rotary_emb": false, | |
| "model.decoder.layers.6.cross_attention.q_proj": false, | |
| "model.decoder.layers.6.cross_attention.k_proj": false, | |
| "model.decoder.layers.6.cross_attention.v_proj": false, | |
| "model.decoder.layers.6.cross_attention.o_proj": false, | |
| "model.decoder.layers.6.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.6.mlp.wi_fused": false, | |
| "model.decoder.layers.6.mlp.activation_fn_0": false, | |
| "model.decoder.layers.6.mlp.activation_fn_1": false, | |
| "model.decoder.layers.6.mlp.dropout": false, | |
| "model.decoder.layers.6.mlp.wo": false, | |
| "model.decoder.layers.7.pre_sa_norm": false, | |
| "model.decoder.layers.7.pre_ca_norm": false, | |
| "model.decoder.layers.7.pre_mlp_norm": false, | |
| "model.decoder.layers.7.self_attention.q_proj": false, | |
| "model.decoder.layers.7.self_attention.k_proj": false, | |
| "model.decoder.layers.7.self_attention.v_proj": false, | |
| "model.decoder.layers.7.self_attention.o_proj": false, | |
| "model.decoder.layers.7.self_attention.rotary_emb": false, | |
| "model.decoder.layers.7.cross_attention.q_proj": false, | |
| "model.decoder.layers.7.cross_attention.k_proj": false, | |
| "model.decoder.layers.7.cross_attention.v_proj": false, | |
| "model.decoder.layers.7.cross_attention.o_proj": false, | |
| "model.decoder.layers.7.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.7.mlp.wi_fused": false, | |
| "model.decoder.layers.7.mlp.activation_fn_0": false, | |
| "model.decoder.layers.7.mlp.activation_fn_1": false, | |
| "model.decoder.layers.7.mlp.dropout": false, | |
| "model.decoder.layers.7.mlp.wo": false, | |
| "model.decoder.layers.8.pre_sa_norm": false, | |
| "model.decoder.layers.8.pre_ca_norm": false, | |
| "model.decoder.layers.8.pre_mlp_norm": false, | |
| "model.decoder.layers.8.self_attention.q_proj": false, | |
| "model.decoder.layers.8.self_attention.k_proj": false, | |
| "model.decoder.layers.8.self_attention.v_proj": false, | |
| "model.decoder.layers.8.self_attention.o_proj": false, | |
| "model.decoder.layers.8.self_attention.rotary_emb": false, | |
| "model.decoder.layers.8.cross_attention.q_proj": false, | |
| "model.decoder.layers.8.cross_attention.k_proj": false, | |
| "model.decoder.layers.8.cross_attention.v_proj": false, | |
| "model.decoder.layers.8.cross_attention.o_proj": false, | |
| "model.decoder.layers.8.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.8.mlp.wi_fused": false, | |
| "model.decoder.layers.8.mlp.activation_fn_0": false, | |
| "model.decoder.layers.8.mlp.activation_fn_1": false, | |
| "model.decoder.layers.8.mlp.dropout": false, | |
| "model.decoder.layers.8.mlp.wo": false, | |
| "model.decoder.layers.9.pre_sa_norm": false, | |
| "model.decoder.layers.9.pre_ca_norm": false, | |
| "model.decoder.layers.9.pre_mlp_norm": false, | |
| "model.decoder.layers.9.self_attention.q_proj": false, | |
| "model.decoder.layers.9.self_attention.k_proj": false, | |
| "model.decoder.layers.9.self_attention.v_proj": false, | |
| "model.decoder.layers.9.self_attention.o_proj": false, | |
| "model.decoder.layers.9.self_attention.rotary_emb": false, | |
| "model.decoder.layers.9.cross_attention.q_proj": false, | |
| "model.decoder.layers.9.cross_attention.k_proj": false, | |
| "model.decoder.layers.9.cross_attention.v_proj": false, | |
| "model.decoder.layers.9.cross_attention.o_proj": false, | |
| "model.decoder.layers.9.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.9.mlp.wi_fused": false, | |
| "model.decoder.layers.9.mlp.activation_fn_0": false, | |
| "model.decoder.layers.9.mlp.activation_fn_1": false, | |
| "model.decoder.layers.9.mlp.dropout": false, | |
| "model.decoder.layers.9.mlp.wo": false, | |
| "model.decoder.layers.10.pre_sa_norm": false, | |
| "model.decoder.layers.10.pre_ca_norm": false, | |
| "model.decoder.layers.10.pre_mlp_norm": false, | |
| "model.decoder.layers.10.self_attention.q_proj": false, | |
| "model.decoder.layers.10.self_attention.k_proj": false, | |
| "model.decoder.layers.10.self_attention.v_proj": false, | |
| "model.decoder.layers.10.self_attention.o_proj": false, | |
| "model.decoder.layers.10.self_attention.rotary_emb": false, | |
| "model.decoder.layers.10.cross_attention.q_proj": false, | |
| "model.decoder.layers.10.cross_attention.k_proj": false, | |
| "model.decoder.layers.10.cross_attention.v_proj": false, | |
| "model.decoder.layers.10.cross_attention.o_proj": false, | |
| "model.decoder.layers.10.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.10.mlp.wi_fused": false, | |
| "model.decoder.layers.10.mlp.activation_fn_0": false, | |
| "model.decoder.layers.10.mlp.activation_fn_1": false, | |
| "model.decoder.layers.10.mlp.dropout": false, | |
| "model.decoder.layers.10.mlp.wo": false, | |
| "model.decoder.layers.11.pre_sa_norm": false, | |
| "model.decoder.layers.11.pre_ca_norm": false, | |
| "model.decoder.layers.11.pre_mlp_norm": false, | |
| "model.decoder.layers.11.self_attention.q_proj": false, | |
| "model.decoder.layers.11.self_attention.k_proj": false, | |
| "model.decoder.layers.11.self_attention.v_proj": false, | |
| "model.decoder.layers.11.self_attention.o_proj": false, | |
| "model.decoder.layers.11.self_attention.rotary_emb": false, | |
| "model.decoder.layers.11.cross_attention.q_proj": false, | |
| "model.decoder.layers.11.cross_attention.k_proj": false, | |
| "model.decoder.layers.11.cross_attention.v_proj": false, | |
| "model.decoder.layers.11.cross_attention.o_proj": false, | |
| "model.decoder.layers.11.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.11.mlp.wi_fused": false, | |
| "model.decoder.layers.11.mlp.activation_fn_0": false, | |
| "model.decoder.layers.11.mlp.activation_fn_1": false, | |
| "model.decoder.layers.11.mlp.dropout": false, | |
| "model.decoder.layers.11.mlp.wo": false, | |
| "model.decoder.layers.12.pre_sa_norm": false, | |
| "model.decoder.layers.12.pre_ca_norm": false, | |
| "model.decoder.layers.12.pre_mlp_norm": false, | |
| "model.decoder.layers.12.self_attention.q_proj": false, | |
| "model.decoder.layers.12.self_attention.k_proj": false, | |
| "model.decoder.layers.12.self_attention.v_proj": false, | |
| "model.decoder.layers.12.self_attention.o_proj": false, | |
| "model.decoder.layers.12.self_attention.rotary_emb": false, | |
| "model.decoder.layers.12.cross_attention.q_proj": false, | |
| "model.decoder.layers.12.cross_attention.k_proj": false, | |
| "model.decoder.layers.12.cross_attention.v_proj": false, | |
| "model.decoder.layers.12.cross_attention.o_proj": false, | |
| "model.decoder.layers.12.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.12.mlp.wi_fused": false, | |
| "model.decoder.layers.12.mlp.activation_fn_0": false, | |
| "model.decoder.layers.12.mlp.activation_fn_1": false, | |
| "model.decoder.layers.12.mlp.dropout": false, | |
| "model.decoder.layers.12.mlp.wo": false, | |
| "model.decoder.layers.13.pre_sa_norm": false, | |
| "model.decoder.layers.13.pre_ca_norm": false, | |
| "model.decoder.layers.13.pre_mlp_norm": false, | |
| "model.decoder.layers.13.self_attention.q_proj": false, | |
| "model.decoder.layers.13.self_attention.k_proj": false, | |
| "model.decoder.layers.13.self_attention.v_proj": false, | |
| "model.decoder.layers.13.self_attention.o_proj": false, | |
| "model.decoder.layers.13.self_attention.rotary_emb": false, | |
| "model.decoder.layers.13.cross_attention.q_proj": false, | |
| "model.decoder.layers.13.cross_attention.k_proj": false, | |
| "model.decoder.layers.13.cross_attention.v_proj": false, | |
| "model.decoder.layers.13.cross_attention.o_proj": false, | |
| "model.decoder.layers.13.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.13.mlp.wi_fused": false, | |
| "model.decoder.layers.13.mlp.activation_fn_0": false, | |
| "model.decoder.layers.13.mlp.activation_fn_1": false, | |
| "model.decoder.layers.13.mlp.dropout": false, | |
| "model.decoder.layers.13.mlp.wo": false, | |
| "model.decoder.layers.14.pre_sa_norm": false, | |
| "model.decoder.layers.14.pre_ca_norm": false, | |
| "model.decoder.layers.14.pre_mlp_norm": false, | |
| "model.decoder.layers.14.self_attention.q_proj": false, | |
| "model.decoder.layers.14.self_attention.k_proj": false, | |
| "model.decoder.layers.14.self_attention.v_proj": false, | |
| "model.decoder.layers.14.self_attention.o_proj": false, | |
| "model.decoder.layers.14.self_attention.rotary_emb": false, | |
| "model.decoder.layers.14.cross_attention.q_proj": false, | |
| "model.decoder.layers.14.cross_attention.k_proj": false, | |
| "model.decoder.layers.14.cross_attention.v_proj": false, | |
| "model.decoder.layers.14.cross_attention.o_proj": false, | |
| "model.decoder.layers.14.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.14.mlp.wi_fused": false, | |
| "model.decoder.layers.14.mlp.activation_fn_0": false, | |
| "model.decoder.layers.14.mlp.activation_fn_1": false, | |
| "model.decoder.layers.14.mlp.dropout": false, | |
| "model.decoder.layers.14.mlp.wo": false, | |
| "model.decoder.layers.15.pre_sa_norm": false, | |
| "model.decoder.layers.15.pre_ca_norm": false, | |
| "model.decoder.layers.15.pre_mlp_norm": false, | |
| "model.decoder.layers.15.self_attention.q_proj": false, | |
| "model.decoder.layers.15.self_attention.k_proj": false, | |
| "model.decoder.layers.15.self_attention.v_proj": false, | |
| "model.decoder.layers.15.self_attention.o_proj": false, | |
| "model.decoder.layers.15.self_attention.rotary_emb": false, | |
| "model.decoder.layers.15.cross_attention.q_proj": false, | |
| "model.decoder.layers.15.cross_attention.k_proj": false, | |
| "model.decoder.layers.15.cross_attention.v_proj": false, | |
| "model.decoder.layers.15.cross_attention.o_proj": false, | |
| "model.decoder.layers.15.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.15.mlp.wi_fused": false, | |
| "model.decoder.layers.15.mlp.activation_fn_0": false, | |
| "model.decoder.layers.15.mlp.activation_fn_1": false, | |
| "model.decoder.layers.15.mlp.dropout": false, | |
| "model.decoder.layers.15.mlp.wo": false, | |
| "model.decoder.layers.16.pre_sa_norm": false, | |
| "model.decoder.layers.16.pre_ca_norm": false, | |
| "model.decoder.layers.16.pre_mlp_norm": false, | |
| "model.decoder.layers.16.self_attention.q_proj": false, | |
| "model.decoder.layers.16.self_attention.k_proj": false, | |
| "model.decoder.layers.16.self_attention.v_proj": false, | |
| "model.decoder.layers.16.self_attention.o_proj": false, | |
| "model.decoder.layers.16.self_attention.rotary_emb": false, | |
| "model.decoder.layers.16.cross_attention.q_proj": false, | |
| "model.decoder.layers.16.cross_attention.k_proj": false, | |
| "model.decoder.layers.16.cross_attention.v_proj": false, | |
| "model.decoder.layers.16.cross_attention.o_proj": false, | |
| "model.decoder.layers.16.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.16.mlp.wi_fused": false, | |
| "model.decoder.layers.16.mlp.activation_fn_0": false, | |
| "model.decoder.layers.16.mlp.activation_fn_1": false, | |
| "model.decoder.layers.16.mlp.dropout": false, | |
| "model.decoder.layers.16.mlp.wo": false, | |
| "model.decoder.layers.17.pre_sa_norm": false, | |
| "model.decoder.layers.17.pre_ca_norm": false, | |
| "model.decoder.layers.17.pre_mlp_norm": false, | |
| "model.decoder.layers.17.self_attention.q_proj": false, | |
| "model.decoder.layers.17.self_attention.k_proj": false, | |
| "model.decoder.layers.17.self_attention.v_proj": false, | |
| "model.decoder.layers.17.self_attention.o_proj": false, | |
| "model.decoder.layers.17.self_attention.rotary_emb": false, | |
| "model.decoder.layers.17.cross_attention.q_proj": false, | |
| "model.decoder.layers.17.cross_attention.k_proj": false, | |
| "model.decoder.layers.17.cross_attention.v_proj": false, | |
| "model.decoder.layers.17.cross_attention.o_proj": false, | |
| "model.decoder.layers.17.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.17.mlp.wi_fused": false, | |
| "model.decoder.layers.17.mlp.activation_fn_0": false, | |
| "model.decoder.layers.17.mlp.activation_fn_1": false, | |
| "model.decoder.layers.17.mlp.dropout": false, | |
| "model.decoder.layers.17.mlp.wo": false, | |
| "model.decoder.norm": false, | |
| "model.decoder.logits_dense": false, | |
| "dac_model.encoder.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.4": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.4": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.4": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.4": false, | |
| "dac_model.encoder.block.layers.5": false, | |
| "dac_model.encoder.block.layers.6": false, | |
| "dac_model.quantizer.quantizers.0.in_proj": false, | |
| "dac_model.quantizer.quantizers.0.out_proj": false, | |
| "dac_model.quantizer.quantizers.0.codebook": false, | |
| "dac_model.quantizer.quantizers.1.in_proj": false, | |
| "dac_model.quantizer.quantizers.1.out_proj": false, | |
| "dac_model.quantizer.quantizers.1.codebook": false, | |
| "dac_model.quantizer.quantizers.2.in_proj": false, | |
| "dac_model.quantizer.quantizers.2.out_proj": false, | |
| "dac_model.quantizer.quantizers.2.codebook": false, | |
| "dac_model.quantizer.quantizers.3.in_proj": false, | |
| "dac_model.quantizer.quantizers.3.out_proj": false, | |
| "dac_model.quantizer.quantizers.3.codebook": false, | |
| "dac_model.quantizer.quantizers.4.in_proj": false, | |
| "dac_model.quantizer.quantizers.4.out_proj": false, | |
| "dac_model.quantizer.quantizers.4.codebook": false, | |
| "dac_model.quantizer.quantizers.5.in_proj": false, | |
| "dac_model.quantizer.quantizers.5.out_proj": false, | |
| "dac_model.quantizer.quantizers.5.codebook": false, | |
| "dac_model.quantizer.quantizers.6.in_proj": false, | |
| "dac_model.quantizer.quantizers.6.out_proj": false, | |
| "dac_model.quantizer.quantizers.6.codebook": false, | |
| "dac_model.quantizer.quantizers.7.in_proj": false, | |
| "dac_model.quantizer.quantizers.7.out_proj": false, | |
| "dac_model.quantizer.quantizers.7.codebook": false, | |
| "dac_model.quantizer.quantizers.8.in_proj": false, | |
| "dac_model.quantizer.quantizers.8.out_proj": false, | |
| "dac_model.quantizer.quantizers.8.codebook": false, | |
| "dac_model.decoder.model.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.5": false, | |
| "dac_model.decoder.model.layers.6": false, | |
| "dac_model.decoder.model.layers.7": false | |
| }, | |
| "quantization_config": { | |
| "group_size": 64, | |
| "bits": 3, | |
| "model.encoder.embedding": false, | |
| "model.encoder.dropout": false, | |
| "model.encoder.layers.0.pre_sa_norm": false, | |
| "model.encoder.layers.0.self_attention.q_proj": false, | |
| "model.encoder.layers.0.self_attention.k_proj": false, | |
| "model.encoder.layers.0.self_attention.v_proj": false, | |
| "model.encoder.layers.0.self_attention.o_proj": false, | |
| "model.encoder.layers.0.self_attention.rotary_emb": false, | |
| "model.encoder.layers.0.post_sa_norm": false, | |
| "model.encoder.layers.0.mlp.wi_fused": false, | |
| "model.encoder.layers.0.mlp.activation_fn_0": false, | |
| "model.encoder.layers.0.mlp.activation_fn_1": false, | |
| "model.encoder.layers.0.mlp.dropout": false, | |
| "model.encoder.layers.0.mlp.wo": false, | |
| "model.encoder.layers.0.dropout": false, | |
| "model.encoder.layers.1.pre_sa_norm": false, | |
| "model.encoder.layers.1.self_attention.q_proj": false, | |
| "model.encoder.layers.1.self_attention.k_proj": false, | |
| "model.encoder.layers.1.self_attention.v_proj": false, | |
| "model.encoder.layers.1.self_attention.o_proj": false, | |
| "model.encoder.layers.1.self_attention.rotary_emb": false, | |
| "model.encoder.layers.1.post_sa_norm": false, | |
| "model.encoder.layers.1.mlp.wi_fused": false, | |
| "model.encoder.layers.1.mlp.activation_fn_0": false, | |
| "model.encoder.layers.1.mlp.activation_fn_1": false, | |
| "model.encoder.layers.1.mlp.dropout": false, | |
| "model.encoder.layers.1.mlp.wo": false, | |
| "model.encoder.layers.1.dropout": false, | |
| "model.encoder.layers.2.pre_sa_norm": false, | |
| "model.encoder.layers.2.self_attention.q_proj": false, | |
| "model.encoder.layers.2.self_attention.k_proj": false, | |
| "model.encoder.layers.2.self_attention.v_proj": false, | |
| "model.encoder.layers.2.self_attention.o_proj": false, | |
| "model.encoder.layers.2.self_attention.rotary_emb": false, | |
| "model.encoder.layers.2.post_sa_norm": false, | |
| "model.encoder.layers.2.mlp.wi_fused": false, | |
| "model.encoder.layers.2.mlp.activation_fn_0": false, | |
| "model.encoder.layers.2.mlp.activation_fn_1": false, | |
| "model.encoder.layers.2.mlp.dropout": false, | |
| "model.encoder.layers.2.mlp.wo": false, | |
| "model.encoder.layers.2.dropout": false, | |
| "model.encoder.layers.3.pre_sa_norm": false, | |
| "model.encoder.layers.3.self_attention.q_proj": false, | |
| "model.encoder.layers.3.self_attention.k_proj": false, | |
| "model.encoder.layers.3.self_attention.v_proj": false, | |
| "model.encoder.layers.3.self_attention.o_proj": false, | |
| "model.encoder.layers.3.self_attention.rotary_emb": false, | |
| "model.encoder.layers.3.post_sa_norm": false, | |
| "model.encoder.layers.3.mlp.wi_fused": false, | |
| "model.encoder.layers.3.mlp.activation_fn_0": false, | |
| "model.encoder.layers.3.mlp.activation_fn_1": false, | |
| "model.encoder.layers.3.mlp.dropout": false, | |
| "model.encoder.layers.3.mlp.wo": false, | |
| "model.encoder.layers.3.dropout": false, | |
| "model.encoder.layers.4.pre_sa_norm": false, | |
| "model.encoder.layers.4.self_attention.q_proj": false, | |
| "model.encoder.layers.4.self_attention.k_proj": false, | |
| "model.encoder.layers.4.self_attention.v_proj": false, | |
| "model.encoder.layers.4.self_attention.o_proj": false, | |
| "model.encoder.layers.4.self_attention.rotary_emb": false, | |
| "model.encoder.layers.4.post_sa_norm": false, | |
| "model.encoder.layers.4.mlp.wi_fused": false, | |
| "model.encoder.layers.4.mlp.activation_fn_0": false, | |
| "model.encoder.layers.4.mlp.activation_fn_1": false, | |
| "model.encoder.layers.4.mlp.dropout": false, | |
| "model.encoder.layers.4.mlp.wo": false, | |
| "model.encoder.layers.4.dropout": false, | |
| "model.encoder.layers.5.pre_sa_norm": false, | |
| "model.encoder.layers.5.self_attention.q_proj": false, | |
| "model.encoder.layers.5.self_attention.k_proj": false, | |
| "model.encoder.layers.5.self_attention.v_proj": false, | |
| "model.encoder.layers.5.self_attention.o_proj": false, | |
| "model.encoder.layers.5.self_attention.rotary_emb": false, | |
| "model.encoder.layers.5.post_sa_norm": false, | |
| "model.encoder.layers.5.mlp.wi_fused": false, | |
| "model.encoder.layers.5.mlp.activation_fn_0": false, | |
| "model.encoder.layers.5.mlp.activation_fn_1": false, | |
| "model.encoder.layers.5.mlp.dropout": false, | |
| "model.encoder.layers.5.mlp.wo": false, | |
| "model.encoder.layers.5.dropout": false, | |
| "model.encoder.layers.6.pre_sa_norm": false, | |
| "model.encoder.layers.6.self_attention.q_proj": false, | |
| "model.encoder.layers.6.self_attention.k_proj": false, | |
| "model.encoder.layers.6.self_attention.v_proj": false, | |
| "model.encoder.layers.6.self_attention.o_proj": false, | |
| "model.encoder.layers.6.self_attention.rotary_emb": false, | |
| "model.encoder.layers.6.post_sa_norm": false, | |
| "model.encoder.layers.6.mlp.wi_fused": false, | |
| "model.encoder.layers.6.mlp.activation_fn_0": false, | |
| "model.encoder.layers.6.mlp.activation_fn_1": false, | |
| "model.encoder.layers.6.mlp.dropout": false, | |
| "model.encoder.layers.6.mlp.wo": false, | |
| "model.encoder.layers.6.dropout": false, | |
| "model.encoder.layers.7.pre_sa_norm": false, | |
| "model.encoder.layers.7.self_attention.q_proj": false, | |
| "model.encoder.layers.7.self_attention.k_proj": false, | |
| "model.encoder.layers.7.self_attention.v_proj": false, | |
| "model.encoder.layers.7.self_attention.o_proj": false, | |
| "model.encoder.layers.7.self_attention.rotary_emb": false, | |
| "model.encoder.layers.7.post_sa_norm": false, | |
| "model.encoder.layers.7.mlp.wi_fused": false, | |
| "model.encoder.layers.7.mlp.activation_fn_0": false, | |
| "model.encoder.layers.7.mlp.activation_fn_1": false, | |
| "model.encoder.layers.7.mlp.dropout": false, | |
| "model.encoder.layers.7.mlp.wo": false, | |
| "model.encoder.layers.7.dropout": false, | |
| "model.encoder.layers.8.pre_sa_norm": false, | |
| "model.encoder.layers.8.self_attention.q_proj": false, | |
| "model.encoder.layers.8.self_attention.k_proj": false, | |
| "model.encoder.layers.8.self_attention.v_proj": false, | |
| "model.encoder.layers.8.self_attention.o_proj": false, | |
| "model.encoder.layers.8.self_attention.rotary_emb": false, | |
| "model.encoder.layers.8.post_sa_norm": false, | |
| "model.encoder.layers.8.mlp.wi_fused": false, | |
| "model.encoder.layers.8.mlp.activation_fn_0": false, | |
| "model.encoder.layers.8.mlp.activation_fn_1": false, | |
| "model.encoder.layers.8.mlp.dropout": false, | |
| "model.encoder.layers.8.mlp.wo": false, | |
| "model.encoder.layers.8.dropout": false, | |
| "model.encoder.layers.9.pre_sa_norm": false, | |
| "model.encoder.layers.9.self_attention.q_proj": false, | |
| "model.encoder.layers.9.self_attention.k_proj": false, | |
| "model.encoder.layers.9.self_attention.v_proj": false, | |
| "model.encoder.layers.9.self_attention.o_proj": false, | |
| "model.encoder.layers.9.self_attention.rotary_emb": false, | |
| "model.encoder.layers.9.post_sa_norm": false, | |
| "model.encoder.layers.9.mlp.wi_fused": false, | |
| "model.encoder.layers.9.mlp.activation_fn_0": false, | |
| "model.encoder.layers.9.mlp.activation_fn_1": false, | |
| "model.encoder.layers.9.mlp.dropout": false, | |
| "model.encoder.layers.9.mlp.wo": false, | |
| "model.encoder.layers.9.dropout": false, | |
| "model.encoder.layers.10.pre_sa_norm": false, | |
| "model.encoder.layers.10.self_attention.q_proj": false, | |
| "model.encoder.layers.10.self_attention.k_proj": false, | |
| "model.encoder.layers.10.self_attention.v_proj": false, | |
| "model.encoder.layers.10.self_attention.o_proj": false, | |
| "model.encoder.layers.10.self_attention.rotary_emb": false, | |
| "model.encoder.layers.10.post_sa_norm": false, | |
| "model.encoder.layers.10.mlp.wi_fused": false, | |
| "model.encoder.layers.10.mlp.activation_fn_0": false, | |
| "model.encoder.layers.10.mlp.activation_fn_1": false, | |
| "model.encoder.layers.10.mlp.dropout": false, | |
| "model.encoder.layers.10.mlp.wo": false, | |
| "model.encoder.layers.10.dropout": false, | |
| "model.encoder.layers.11.pre_sa_norm": false, | |
| "model.encoder.layers.11.self_attention.q_proj": false, | |
| "model.encoder.layers.11.self_attention.k_proj": false, | |
| "model.encoder.layers.11.self_attention.v_proj": false, | |
| "model.encoder.layers.11.self_attention.o_proj": false, | |
| "model.encoder.layers.11.self_attention.rotary_emb": false, | |
| "model.encoder.layers.11.post_sa_norm": false, | |
| "model.encoder.layers.11.mlp.wi_fused": false, | |
| "model.encoder.layers.11.mlp.activation_fn_0": false, | |
| "model.encoder.layers.11.mlp.activation_fn_1": false, | |
| "model.encoder.layers.11.mlp.dropout": false, | |
| "model.encoder.layers.11.mlp.wo": false, | |
| "model.encoder.layers.11.dropout": false, | |
| "model.encoder.norm": false, | |
| "model.decoder.embeddings.0": false, | |
| "model.decoder.embeddings.1": false, | |
| "model.decoder.embeddings.2": false, | |
| "model.decoder.embeddings.3": false, | |
| "model.decoder.embeddings.4": false, | |
| "model.decoder.embeddings.5": false, | |
| "model.decoder.embeddings.6": false, | |
| "model.decoder.embeddings.7": false, | |
| "model.decoder.embeddings.8": false, | |
| "model.decoder.dropout": false, | |
| "model.decoder.layers.0.pre_sa_norm": false, | |
| "model.decoder.layers.0.pre_ca_norm": false, | |
| "model.decoder.layers.0.pre_mlp_norm": false, | |
| "model.decoder.layers.0.self_attention.q_proj": false, | |
| "model.decoder.layers.0.self_attention.k_proj": false, | |
| "model.decoder.layers.0.self_attention.v_proj": false, | |
| "model.decoder.layers.0.self_attention.o_proj": false, | |
| "model.decoder.layers.0.self_attention.rotary_emb": false, | |
| "model.decoder.layers.0.cross_attention.q_proj": false, | |
| "model.decoder.layers.0.cross_attention.k_proj": false, | |
| "model.decoder.layers.0.cross_attention.v_proj": false, | |
| "model.decoder.layers.0.cross_attention.o_proj": false, | |
| "model.decoder.layers.0.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.0.mlp.wi_fused": false, | |
| "model.decoder.layers.0.mlp.activation_fn_0": false, | |
| "model.decoder.layers.0.mlp.activation_fn_1": false, | |
| "model.decoder.layers.0.mlp.dropout": false, | |
| "model.decoder.layers.0.mlp.wo": false, | |
| "model.decoder.layers.1.pre_sa_norm": false, | |
| "model.decoder.layers.1.pre_ca_norm": false, | |
| "model.decoder.layers.1.pre_mlp_norm": false, | |
| "model.decoder.layers.1.self_attention.q_proj": false, | |
| "model.decoder.layers.1.self_attention.k_proj": false, | |
| "model.decoder.layers.1.self_attention.v_proj": false, | |
| "model.decoder.layers.1.self_attention.o_proj": false, | |
| "model.decoder.layers.1.self_attention.rotary_emb": false, | |
| "model.decoder.layers.1.cross_attention.q_proj": false, | |
| "model.decoder.layers.1.cross_attention.k_proj": false, | |
| "model.decoder.layers.1.cross_attention.v_proj": false, | |
| "model.decoder.layers.1.cross_attention.o_proj": false, | |
| "model.decoder.layers.1.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.1.mlp.wi_fused": false, | |
| "model.decoder.layers.1.mlp.activation_fn_0": false, | |
| "model.decoder.layers.1.mlp.activation_fn_1": false, | |
| "model.decoder.layers.1.mlp.dropout": false, | |
| "model.decoder.layers.1.mlp.wo": false, | |
| "model.decoder.layers.2.pre_sa_norm": false, | |
| "model.decoder.layers.2.pre_ca_norm": false, | |
| "model.decoder.layers.2.pre_mlp_norm": false, | |
| "model.decoder.layers.2.self_attention.q_proj": false, | |
| "model.decoder.layers.2.self_attention.k_proj": false, | |
| "model.decoder.layers.2.self_attention.v_proj": false, | |
| "model.decoder.layers.2.self_attention.o_proj": false, | |
| "model.decoder.layers.2.self_attention.rotary_emb": false, | |
| "model.decoder.layers.2.cross_attention.q_proj": false, | |
| "model.decoder.layers.2.cross_attention.k_proj": false, | |
| "model.decoder.layers.2.cross_attention.v_proj": false, | |
| "model.decoder.layers.2.cross_attention.o_proj": false, | |
| "model.decoder.layers.2.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.2.mlp.wi_fused": false, | |
| "model.decoder.layers.2.mlp.activation_fn_0": false, | |
| "model.decoder.layers.2.mlp.activation_fn_1": false, | |
| "model.decoder.layers.2.mlp.dropout": false, | |
| "model.decoder.layers.2.mlp.wo": false, | |
| "model.decoder.layers.3.pre_sa_norm": false, | |
| "model.decoder.layers.3.pre_ca_norm": false, | |
| "model.decoder.layers.3.pre_mlp_norm": false, | |
| "model.decoder.layers.3.self_attention.q_proj": false, | |
| "model.decoder.layers.3.self_attention.k_proj": false, | |
| "model.decoder.layers.3.self_attention.v_proj": false, | |
| "model.decoder.layers.3.self_attention.o_proj": false, | |
| "model.decoder.layers.3.self_attention.rotary_emb": false, | |
| "model.decoder.layers.3.cross_attention.q_proj": false, | |
| "model.decoder.layers.3.cross_attention.k_proj": false, | |
| "model.decoder.layers.3.cross_attention.v_proj": false, | |
| "model.decoder.layers.3.cross_attention.o_proj": false, | |
| "model.decoder.layers.3.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.3.mlp.wi_fused": false, | |
| "model.decoder.layers.3.mlp.activation_fn_0": false, | |
| "model.decoder.layers.3.mlp.activation_fn_1": false, | |
| "model.decoder.layers.3.mlp.dropout": false, | |
| "model.decoder.layers.3.mlp.wo": false, | |
| "model.decoder.layers.4.pre_sa_norm": false, | |
| "model.decoder.layers.4.pre_ca_norm": false, | |
| "model.decoder.layers.4.pre_mlp_norm": false, | |
| "model.decoder.layers.4.self_attention.q_proj": false, | |
| "model.decoder.layers.4.self_attention.k_proj": false, | |
| "model.decoder.layers.4.self_attention.v_proj": false, | |
| "model.decoder.layers.4.self_attention.o_proj": false, | |
| "model.decoder.layers.4.self_attention.rotary_emb": false, | |
| "model.decoder.layers.4.cross_attention.q_proj": false, | |
| "model.decoder.layers.4.cross_attention.k_proj": false, | |
| "model.decoder.layers.4.cross_attention.v_proj": false, | |
| "model.decoder.layers.4.cross_attention.o_proj": false, | |
| "model.decoder.layers.4.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.4.mlp.wi_fused": false, | |
| "model.decoder.layers.4.mlp.activation_fn_0": false, | |
| "model.decoder.layers.4.mlp.activation_fn_1": false, | |
| "model.decoder.layers.4.mlp.dropout": false, | |
| "model.decoder.layers.4.mlp.wo": false, | |
| "model.decoder.layers.5.pre_sa_norm": false, | |
| "model.decoder.layers.5.pre_ca_norm": false, | |
| "model.decoder.layers.5.pre_mlp_norm": false, | |
| "model.decoder.layers.5.self_attention.q_proj": false, | |
| "model.decoder.layers.5.self_attention.k_proj": false, | |
| "model.decoder.layers.5.self_attention.v_proj": false, | |
| "model.decoder.layers.5.self_attention.o_proj": false, | |
| "model.decoder.layers.5.self_attention.rotary_emb": false, | |
| "model.decoder.layers.5.cross_attention.q_proj": false, | |
| "model.decoder.layers.5.cross_attention.k_proj": false, | |
| "model.decoder.layers.5.cross_attention.v_proj": false, | |
| "model.decoder.layers.5.cross_attention.o_proj": false, | |
| "model.decoder.layers.5.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.5.mlp.wi_fused": false, | |
| "model.decoder.layers.5.mlp.activation_fn_0": false, | |
| "model.decoder.layers.5.mlp.activation_fn_1": false, | |
| "model.decoder.layers.5.mlp.dropout": false, | |
| "model.decoder.layers.5.mlp.wo": false, | |
| "model.decoder.layers.6.pre_sa_norm": false, | |
| "model.decoder.layers.6.pre_ca_norm": false, | |
| "model.decoder.layers.6.pre_mlp_norm": false, | |
| "model.decoder.layers.6.self_attention.q_proj": false, | |
| "model.decoder.layers.6.self_attention.k_proj": false, | |
| "model.decoder.layers.6.self_attention.v_proj": false, | |
| "model.decoder.layers.6.self_attention.o_proj": false, | |
| "model.decoder.layers.6.self_attention.rotary_emb": false, | |
| "model.decoder.layers.6.cross_attention.q_proj": false, | |
| "model.decoder.layers.6.cross_attention.k_proj": false, | |
| "model.decoder.layers.6.cross_attention.v_proj": false, | |
| "model.decoder.layers.6.cross_attention.o_proj": false, | |
| "model.decoder.layers.6.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.6.mlp.wi_fused": false, | |
| "model.decoder.layers.6.mlp.activation_fn_0": false, | |
| "model.decoder.layers.6.mlp.activation_fn_1": false, | |
| "model.decoder.layers.6.mlp.dropout": false, | |
| "model.decoder.layers.6.mlp.wo": false, | |
| "model.decoder.layers.7.pre_sa_norm": false, | |
| "model.decoder.layers.7.pre_ca_norm": false, | |
| "model.decoder.layers.7.pre_mlp_norm": false, | |
| "model.decoder.layers.7.self_attention.q_proj": false, | |
| "model.decoder.layers.7.self_attention.k_proj": false, | |
| "model.decoder.layers.7.self_attention.v_proj": false, | |
| "model.decoder.layers.7.self_attention.o_proj": false, | |
| "model.decoder.layers.7.self_attention.rotary_emb": false, | |
| "model.decoder.layers.7.cross_attention.q_proj": false, | |
| "model.decoder.layers.7.cross_attention.k_proj": false, | |
| "model.decoder.layers.7.cross_attention.v_proj": false, | |
| "model.decoder.layers.7.cross_attention.o_proj": false, | |
| "model.decoder.layers.7.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.7.mlp.wi_fused": false, | |
| "model.decoder.layers.7.mlp.activation_fn_0": false, | |
| "model.decoder.layers.7.mlp.activation_fn_1": false, | |
| "model.decoder.layers.7.mlp.dropout": false, | |
| "model.decoder.layers.7.mlp.wo": false, | |
| "model.decoder.layers.8.pre_sa_norm": false, | |
| "model.decoder.layers.8.pre_ca_norm": false, | |
| "model.decoder.layers.8.pre_mlp_norm": false, | |
| "model.decoder.layers.8.self_attention.q_proj": false, | |
| "model.decoder.layers.8.self_attention.k_proj": false, | |
| "model.decoder.layers.8.self_attention.v_proj": false, | |
| "model.decoder.layers.8.self_attention.o_proj": false, | |
| "model.decoder.layers.8.self_attention.rotary_emb": false, | |
| "model.decoder.layers.8.cross_attention.q_proj": false, | |
| "model.decoder.layers.8.cross_attention.k_proj": false, | |
| "model.decoder.layers.8.cross_attention.v_proj": false, | |
| "model.decoder.layers.8.cross_attention.o_proj": false, | |
| "model.decoder.layers.8.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.8.mlp.wi_fused": false, | |
| "model.decoder.layers.8.mlp.activation_fn_0": false, | |
| "model.decoder.layers.8.mlp.activation_fn_1": false, | |
| "model.decoder.layers.8.mlp.dropout": false, | |
| "model.decoder.layers.8.mlp.wo": false, | |
| "model.decoder.layers.9.pre_sa_norm": false, | |
| "model.decoder.layers.9.pre_ca_norm": false, | |
| "model.decoder.layers.9.pre_mlp_norm": false, | |
| "model.decoder.layers.9.self_attention.q_proj": false, | |
| "model.decoder.layers.9.self_attention.k_proj": false, | |
| "model.decoder.layers.9.self_attention.v_proj": false, | |
| "model.decoder.layers.9.self_attention.o_proj": false, | |
| "model.decoder.layers.9.self_attention.rotary_emb": false, | |
| "model.decoder.layers.9.cross_attention.q_proj": false, | |
| "model.decoder.layers.9.cross_attention.k_proj": false, | |
| "model.decoder.layers.9.cross_attention.v_proj": false, | |
| "model.decoder.layers.9.cross_attention.o_proj": false, | |
| "model.decoder.layers.9.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.9.mlp.wi_fused": false, | |
| "model.decoder.layers.9.mlp.activation_fn_0": false, | |
| "model.decoder.layers.9.mlp.activation_fn_1": false, | |
| "model.decoder.layers.9.mlp.dropout": false, | |
| "model.decoder.layers.9.mlp.wo": false, | |
| "model.decoder.layers.10.pre_sa_norm": false, | |
| "model.decoder.layers.10.pre_ca_norm": false, | |
| "model.decoder.layers.10.pre_mlp_norm": false, | |
| "model.decoder.layers.10.self_attention.q_proj": false, | |
| "model.decoder.layers.10.self_attention.k_proj": false, | |
| "model.decoder.layers.10.self_attention.v_proj": false, | |
| "model.decoder.layers.10.self_attention.o_proj": false, | |
| "model.decoder.layers.10.self_attention.rotary_emb": false, | |
| "model.decoder.layers.10.cross_attention.q_proj": false, | |
| "model.decoder.layers.10.cross_attention.k_proj": false, | |
| "model.decoder.layers.10.cross_attention.v_proj": false, | |
| "model.decoder.layers.10.cross_attention.o_proj": false, | |
| "model.decoder.layers.10.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.10.mlp.wi_fused": false, | |
| "model.decoder.layers.10.mlp.activation_fn_0": false, | |
| "model.decoder.layers.10.mlp.activation_fn_1": false, | |
| "model.decoder.layers.10.mlp.dropout": false, | |
| "model.decoder.layers.10.mlp.wo": false, | |
| "model.decoder.layers.11.pre_sa_norm": false, | |
| "model.decoder.layers.11.pre_ca_norm": false, | |
| "model.decoder.layers.11.pre_mlp_norm": false, | |
| "model.decoder.layers.11.self_attention.q_proj": false, | |
| "model.decoder.layers.11.self_attention.k_proj": false, | |
| "model.decoder.layers.11.self_attention.v_proj": false, | |
| "model.decoder.layers.11.self_attention.o_proj": false, | |
| "model.decoder.layers.11.self_attention.rotary_emb": false, | |
| "model.decoder.layers.11.cross_attention.q_proj": false, | |
| "model.decoder.layers.11.cross_attention.k_proj": false, | |
| "model.decoder.layers.11.cross_attention.v_proj": false, | |
| "model.decoder.layers.11.cross_attention.o_proj": false, | |
| "model.decoder.layers.11.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.11.mlp.wi_fused": false, | |
| "model.decoder.layers.11.mlp.activation_fn_0": false, | |
| "model.decoder.layers.11.mlp.activation_fn_1": false, | |
| "model.decoder.layers.11.mlp.dropout": false, | |
| "model.decoder.layers.11.mlp.wo": false, | |
| "model.decoder.layers.12.pre_sa_norm": false, | |
| "model.decoder.layers.12.pre_ca_norm": false, | |
| "model.decoder.layers.12.pre_mlp_norm": false, | |
| "model.decoder.layers.12.self_attention.q_proj": false, | |
| "model.decoder.layers.12.self_attention.k_proj": false, | |
| "model.decoder.layers.12.self_attention.v_proj": false, | |
| "model.decoder.layers.12.self_attention.o_proj": false, | |
| "model.decoder.layers.12.self_attention.rotary_emb": false, | |
| "model.decoder.layers.12.cross_attention.q_proj": false, | |
| "model.decoder.layers.12.cross_attention.k_proj": false, | |
| "model.decoder.layers.12.cross_attention.v_proj": false, | |
| "model.decoder.layers.12.cross_attention.o_proj": false, | |
| "model.decoder.layers.12.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.12.mlp.wi_fused": false, | |
| "model.decoder.layers.12.mlp.activation_fn_0": false, | |
| "model.decoder.layers.12.mlp.activation_fn_1": false, | |
| "model.decoder.layers.12.mlp.dropout": false, | |
| "model.decoder.layers.12.mlp.wo": false, | |
| "model.decoder.layers.13.pre_sa_norm": false, | |
| "model.decoder.layers.13.pre_ca_norm": false, | |
| "model.decoder.layers.13.pre_mlp_norm": false, | |
| "model.decoder.layers.13.self_attention.q_proj": false, | |
| "model.decoder.layers.13.self_attention.k_proj": false, | |
| "model.decoder.layers.13.self_attention.v_proj": false, | |
| "model.decoder.layers.13.self_attention.o_proj": false, | |
| "model.decoder.layers.13.self_attention.rotary_emb": false, | |
| "model.decoder.layers.13.cross_attention.q_proj": false, | |
| "model.decoder.layers.13.cross_attention.k_proj": false, | |
| "model.decoder.layers.13.cross_attention.v_proj": false, | |
| "model.decoder.layers.13.cross_attention.o_proj": false, | |
| "model.decoder.layers.13.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.13.mlp.wi_fused": false, | |
| "model.decoder.layers.13.mlp.activation_fn_0": false, | |
| "model.decoder.layers.13.mlp.activation_fn_1": false, | |
| "model.decoder.layers.13.mlp.dropout": false, | |
| "model.decoder.layers.13.mlp.wo": false, | |
| "model.decoder.layers.14.pre_sa_norm": false, | |
| "model.decoder.layers.14.pre_ca_norm": false, | |
| "model.decoder.layers.14.pre_mlp_norm": false, | |
| "model.decoder.layers.14.self_attention.q_proj": false, | |
| "model.decoder.layers.14.self_attention.k_proj": false, | |
| "model.decoder.layers.14.self_attention.v_proj": false, | |
| "model.decoder.layers.14.self_attention.o_proj": false, | |
| "model.decoder.layers.14.self_attention.rotary_emb": false, | |
| "model.decoder.layers.14.cross_attention.q_proj": false, | |
| "model.decoder.layers.14.cross_attention.k_proj": false, | |
| "model.decoder.layers.14.cross_attention.v_proj": false, | |
| "model.decoder.layers.14.cross_attention.o_proj": false, | |
| "model.decoder.layers.14.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.14.mlp.wi_fused": false, | |
| "model.decoder.layers.14.mlp.activation_fn_0": false, | |
| "model.decoder.layers.14.mlp.activation_fn_1": false, | |
| "model.decoder.layers.14.mlp.dropout": false, | |
| "model.decoder.layers.14.mlp.wo": false, | |
| "model.decoder.layers.15.pre_sa_norm": false, | |
| "model.decoder.layers.15.pre_ca_norm": false, | |
| "model.decoder.layers.15.pre_mlp_norm": false, | |
| "model.decoder.layers.15.self_attention.q_proj": false, | |
| "model.decoder.layers.15.self_attention.k_proj": false, | |
| "model.decoder.layers.15.self_attention.v_proj": false, | |
| "model.decoder.layers.15.self_attention.o_proj": false, | |
| "model.decoder.layers.15.self_attention.rotary_emb": false, | |
| "model.decoder.layers.15.cross_attention.q_proj": false, | |
| "model.decoder.layers.15.cross_attention.k_proj": false, | |
| "model.decoder.layers.15.cross_attention.v_proj": false, | |
| "model.decoder.layers.15.cross_attention.o_proj": false, | |
| "model.decoder.layers.15.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.15.mlp.wi_fused": false, | |
| "model.decoder.layers.15.mlp.activation_fn_0": false, | |
| "model.decoder.layers.15.mlp.activation_fn_1": false, | |
| "model.decoder.layers.15.mlp.dropout": false, | |
| "model.decoder.layers.15.mlp.wo": false, | |
| "model.decoder.layers.16.pre_sa_norm": false, | |
| "model.decoder.layers.16.pre_ca_norm": false, | |
| "model.decoder.layers.16.pre_mlp_norm": false, | |
| "model.decoder.layers.16.self_attention.q_proj": false, | |
| "model.decoder.layers.16.self_attention.k_proj": false, | |
| "model.decoder.layers.16.self_attention.v_proj": false, | |
| "model.decoder.layers.16.self_attention.o_proj": false, | |
| "model.decoder.layers.16.self_attention.rotary_emb": false, | |
| "model.decoder.layers.16.cross_attention.q_proj": false, | |
| "model.decoder.layers.16.cross_attention.k_proj": false, | |
| "model.decoder.layers.16.cross_attention.v_proj": false, | |
| "model.decoder.layers.16.cross_attention.o_proj": false, | |
| "model.decoder.layers.16.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.16.mlp.wi_fused": false, | |
| "model.decoder.layers.16.mlp.activation_fn_0": false, | |
| "model.decoder.layers.16.mlp.activation_fn_1": false, | |
| "model.decoder.layers.16.mlp.dropout": false, | |
| "model.decoder.layers.16.mlp.wo": false, | |
| "model.decoder.layers.17.pre_sa_norm": false, | |
| "model.decoder.layers.17.pre_ca_norm": false, | |
| "model.decoder.layers.17.pre_mlp_norm": false, | |
| "model.decoder.layers.17.self_attention.q_proj": false, | |
| "model.decoder.layers.17.self_attention.k_proj": false, | |
| "model.decoder.layers.17.self_attention.v_proj": false, | |
| "model.decoder.layers.17.self_attention.o_proj": false, | |
| "model.decoder.layers.17.self_attention.rotary_emb": false, | |
| "model.decoder.layers.17.cross_attention.q_proj": false, | |
| "model.decoder.layers.17.cross_attention.k_proj": false, | |
| "model.decoder.layers.17.cross_attention.v_proj": false, | |
| "model.decoder.layers.17.cross_attention.o_proj": false, | |
| "model.decoder.layers.17.cross_attention.rotary_emb": false, | |
| "model.decoder.layers.17.mlp.wi_fused": false, | |
| "model.decoder.layers.17.mlp.activation_fn_0": false, | |
| "model.decoder.layers.17.mlp.activation_fn_1": false, | |
| "model.decoder.layers.17.mlp.dropout": false, | |
| "model.decoder.layers.17.mlp.wo": false, | |
| "model.decoder.norm": false, | |
| "model.decoder.logits_dense": false, | |
| "dac_model.encoder.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.1.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.1.block.layers.4": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.2.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.2.block.layers.4": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.3.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.3": false, | |
| "dac_model.encoder.block.layers.3.block.layers.4": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.0.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.1.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.0": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.1": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.2": false, | |
| "dac_model.encoder.block.layers.4.block.layers.2.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.3": false, | |
| "dac_model.encoder.block.layers.4.block.layers.4": false, | |
| "dac_model.encoder.block.layers.5": false, | |
| "dac_model.encoder.block.layers.6": false, | |
| "dac_model.quantizer.quantizers.0.in_proj": false, | |
| "dac_model.quantizer.quantizers.0.out_proj": false, | |
| "dac_model.quantizer.quantizers.0.codebook": false, | |
| "dac_model.quantizer.quantizers.1.in_proj": false, | |
| "dac_model.quantizer.quantizers.1.out_proj": false, | |
| "dac_model.quantizer.quantizers.1.codebook": false, | |
| "dac_model.quantizer.quantizers.2.in_proj": false, | |
| "dac_model.quantizer.quantizers.2.out_proj": false, | |
| "dac_model.quantizer.quantizers.2.codebook": false, | |
| "dac_model.quantizer.quantizers.3.in_proj": false, | |
| "dac_model.quantizer.quantizers.3.out_proj": false, | |
| "dac_model.quantizer.quantizers.3.codebook": false, | |
| "dac_model.quantizer.quantizers.4.in_proj": false, | |
| "dac_model.quantizer.quantizers.4.out_proj": false, | |
| "dac_model.quantizer.quantizers.4.codebook": false, | |
| "dac_model.quantizer.quantizers.5.in_proj": false, | |
| "dac_model.quantizer.quantizers.5.out_proj": false, | |
| "dac_model.quantizer.quantizers.5.codebook": false, | |
| "dac_model.quantizer.quantizers.6.in_proj": false, | |
| "dac_model.quantizer.quantizers.6.out_proj": false, | |
| "dac_model.quantizer.quantizers.6.codebook": false, | |
| "dac_model.quantizer.quantizers.7.in_proj": false, | |
| "dac_model.quantizer.quantizers.7.out_proj": false, | |
| "dac_model.quantizer.quantizers.7.codebook": false, | |
| "dac_model.quantizer.quantizers.8.in_proj": false, | |
| "dac_model.quantizer.quantizers.8.out_proj": false, | |
| "dac_model.quantizer.quantizers.8.codebook": false, | |
| "dac_model.decoder.model.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.1.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.2.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.3.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.2.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.3.block.layers.3": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.0": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.1": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.2": false, | |
| "dac_model.decoder.model.layers.4.block.layers.4.block.layers.3": false, | |
| "dac_model.decoder.model.layers.5": false, | |
| "dac_model.decoder.model.layers.6": false, | |
| "dac_model.decoder.model.layers.7": false | |
| }, | |
| "training": {}, | |
| "version": "0.1" | |
| } |