NeoLLM / config.json
KitsuVp's picture
Model save
7ff57b3 verified
Raw
History Blame Contribute Delete
4.09 kB
{
"affine_momentum": 0.9,
"architectures": [
"NeoLLMForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.1,
"attn_res_num_blocks": 4,
"auto_map": {
"AutoConfig": "configuration_neollm.NeoLLMConfig",
"AutoModel": "modeling_neollm.NeoLLMModel",
"AutoModelForCausalLM": "modeling_neollm.NeoLLMForCausalLM"
},
"bos_token_id": 1,
"cce_loss_impl": "cce_kahan_full_c",
"directional_routing_k": 4,
"directional_routing_temp": 3.0,
"dropout_rate": 0.1,
"dtype": "bfloat16",
"eos_token_id": 7,
"fan_ratio": 0.125,
"fan_ratio_ffn": 0.0625,
"generator_d_seed": 128,
"generator_k": 3,
"generator_krank": 64,
"generator_num_knots": 32,
"generator_num_modes": 8,
"generator_spline_degree": 2,
"head_dim": 64,
"hidden_act": "xielu",
"hidden_size": 512,
"hola_chunk_size": 256,
"hola_logit_bias_init": -2.0,
"hola_memory_size": 64,
"hola_rms_gamma_init": 1.0,
"hola_score_eps": 1e-06,
"iha_global_layers_use_iha": true,
"iha_local_global_pattern": "LLLLG",
"iha_num_pseudo_heads": 2,
"iha_sliding_window": null,
"initializer_range": 0.02,
"intermediate_size": 1536,
"jtokm_aux_loss_weight": 0.0001,
"jtokm_norm_eps": 1e-06,
"jtokm_num_experts": 4,
"jtokm_num_modes": 4,
"jtokm_top_k": 2,
"laurel_lr_rank": 32,
"liger_loss_accum_dtype": "float32",
"lucid_attention_eps": 1e-06,
"max_position_embeddings": 512,
"mea_component_key_value_heads": 4,
"mea_groupnorm_eps": 1e-06,
"meap_exclude_last": true,
"meap_implementation": "triton",
"meap_mask_ratio": 0.15,
"meap_mask_token_id": 0,
"meap_seed": 42,
"mile_loss_gamma": 1.0,
"model_type": "neollm",
"momentum_gamma": 0.1,
"mu_loss_lambda": 0.0001,
"nextlat_dynamics_proj_factor": 1.0,
"nextlat_horizon": 1,
"nextlat_kl_weight": 1.0,
"nextlat_mse_weight": 1.0,
"nitp_loss_weight": 1.0,
"nitp_projector_intermediate_size": 2048,
"nitp_target_layer": 2,
"nitp_temporal_apply_loss": true,
"nitp_temporal_dynamics_weight": 1.0,
"nitp_temporal_horizon": 4,
"nitp_temporal_identification_weight": 0.1,
"nitp_temporal_intermediate_size": 1024,
"nitp_temporal_similarity_temperature": 0.1,
"nitp_temporal_target_temperature": 0.5,
"ntp_loss_backend": "cce",
"num_attention_heads": 8,
"num_hidden_layers": 12,
"num_key_value_heads": 4,
"num_mem_heads": 4,
"pad_token_id": 0,
"partial_rotary_factor": 0.25,
"polynorm_exclusive": false,
"repo_d_p": 64,
"repo_goat_num_frequencies": 3,
"repo_goat_sink_decay": 4.0,
"repo_start_layer": 4,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"partial_rotary_factor": 0.25,
"rope_theta": 10000.0,
"rope_type": "default"
},
"rope_theta": 10000.0,
"siamese_attn_x_scale_init": 1.0,
"siamese_depth_scaling": true,
"siamese_normalized_input": true,
"stack_d_model": 24,
"stack_memory_cache_size": 2048,
"stack_slots": 16,
"tie_word_embeddings": false,
"transformers_version": "5.14.1",
"tweo_eps": 1e-06,
"tweo_loss_weight": 0.01,
"tweo_power": 4.0,
"tweo_tau": 3.0,
"use_affine_scaled_attention": false,
"use_attn_res": false,
"use_cache": false,
"use_directional_routing": false,
"use_embedding_input_norm": true,
"use_embedding_multipliers": false,
"use_fan_residual": false,
"use_gpas": false,
"use_hadamard_o_proj": true,
"use_hola_memory": false,
"use_iha": false,
"use_jtokm": false,
"use_laurel": false,
"use_laurel_lr": false,
"use_laurel_rw": false,
"use_learnable_multipliers": true,
"use_liger_kernel": false,
"use_lns": true,
"use_lucid_attention": false,
"use_mea_attention": false,
"use_meap": true,
"use_mile_loss": true,
"use_momentum_attention": true,
"use_mu_loss": true,
"use_nextlat": false,
"use_nitp": true,
"use_nitp_temporal": false,
"use_repo": true,
"use_repo_goat_prior": false,
"use_repo_grape": true,
"use_siamesenorm": true,
"use_spelling_bee_embeddings": true,
"use_stack_memory": false,
"use_token_generator": false,
"use_tweo": true,
"use_xsa": true,
"vocab_size": 64402,
"xsa_eps": 1e-06
}