| { |
| "allow_negative_eigenvalues": false, |
| "architecture": "Senryu2ForCausalLM", |
| "attention_dropout": 0.0, |
| "attention_head_dim": 128, |
| "attention_layer_type": "parallax", |
| "bos_token_id": 0, |
| "eos_token_id": 1, |
| "full_attention_interval": 4, |
| "hidden_size": 1536, |
| "initializer_range": 0.02, |
| "intermediate_size": 3584, |
| "kda_lower_bound": -5.0, |
| "kda_safe_gate": true, |
| "layer_types": [ |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax", |
| "gdn2", |
| "gdn2", |
| "gdn2", |
| "parallax" |
| ], |
| "linear_conv_kernel_dim": 4, |
| "linear_expand_v": 1.0, |
| "linear_full_rank_gate": true, |
| "linear_head_dim": 128, |
| "linear_layer_type": "gdn2", |
| "linear_num_heads": 12, |
| "linear_num_value_heads": 12, |
| "max_position_embeddings": 32768, |
| "mla_kv_lora_rank": 256, |
| "mla_q_lora_rank": 512, |
| "mla_qk_nope_head_dim": 128, |
| "mla_v_head_dim": 128, |
| "mlp_activation": "situ_glu", |
| "model_type": "senryu2", |
| "mtp_adapter_rank": 128, |
| "mtp_loss_weight": 0.0, |
| "mtp_num_heads": 0, |
| "ngram_conv_kernel": 4, |
| "ngram_embedding_dim": 128, |
| "ngram_heads_per_order": 4, |
| "ngram_layer_index": 1, |
| "ngram_orders": [ |
| 2, |
| 3 |
| ], |
| "ngram_table_rows": 2097152, |
| "num_attention_heads": 12, |
| "num_hidden_layers": 28, |
| "num_key_value_heads": 4, |
| "pad_token_id": 2, |
| "partial_rotary_factor": 0.5, |
| "residual_branches": 4, |
| "residual_gate_rank": 128, |
| "residual_mode": "gated_residual", |
| "rms_norm_eps": 1e-06, |
| "rope_theta": 1000000.0, |
| "shared_layer_indices": [], |
| "situ_gate_cap": 4.0, |
| "situ_up_cap": 25.0, |
| "tie_word_embeddings": true, |
| "training_sequence_length": 32768, |
| "unk_token_id": 3, |
| "use_block_attn_res": false, |
| "vocab_size": 65536, |
| "architectures": [ |
| "Senryu2ForCausalLM" |
| ], |
| "torch_dtype": "bfloat16", |
| "model_name": "Senryu-2-1B", |
| "model_family": "Senryu-2", |
| "developer": "KerzgrAI", |
| "weights": "ema", |
| "core_parameters": 1094961660, |
| "ngram_table_parameters": 2166398464, |
| "chat_template": "smollm3", |
| "reasoning": true, |
| "tool_calls": "hermes_json" |
| } |
|
|
