{ "architectures": ["TweaktronOmniForCausalLM"], "model_type": "custom_model", "auto_map": { "AutoConfig": "configuration_tweaktron.TweaktronOmniConfig", "AutoModel": "modeling_tweaktron.TweaktronOmniModel", "AutoModelForCausalLM": "modeling_tweaktron.TweaktronOmniForCausalLM" }, "hidden_size": 2048, "num_heads": 14, "num_hidden_layers": 3, "hidden_ratio": 4, "intermediate_size": null, "hidden_act": "swish", "norm_eps": 1e-06, "max_position_embeddings": 4096, "comba_num_heads": 6, "comba_head_dim": 256, "comba_expand_v": 2.0, "comba_use_output_gate": true, "comba_use_output_correction": true, "comba_use_inner_decay": true, "comba_correction_factor": 1.0, "comba_conv_size": 4, "mom_num_heads": 4, "mom_head_dim": 256, "mom_expand_v": 1.0, "mom_num_memories": 8, "mom_topk": 2, "mom_capacity": 1.0, "mom_shared_mem": true, "mom_single_kv_proj": false, "mom_use_output_gate": true, "mom_conv_size": 4, "aux_loss_scale": 0.01, "raven_num_heads": 4, "raven_num_kv_heads": null, "raven_num_slots": 64, "raven_expand_k": 1.0, "raven_expand_v": 1.0, "raven_feature_map": "swish", "raven_use_output_gate": false, "raven_decay_type": "Mamba2", "raven_topk": 32, "raven_bias_rmm": false, "raven_add_gumbel_noise": true, "raven_router_score": "sigmoid", "raven_router_type": "lin", "raven_gate_logit_normalizer": 8, "vocab_size": 32016, "bos_token_id": 1, "eos_token_id": 2, "pad_token_id": null, "tie_word_embeddings": false, "initializer_range": 0.02, "use_cache": true, "fuse_norm": true, "fuse_swiglu": true, "fuse_cross_entropy": true, "fuse_linear_cross_entropy": false, "torch_dtype": "bfloat16", "transformers_version": "5.0.0" }