| { |
| "architectures": [ |
| "MarkV3ForCausalLM" |
| ], |
| "auto_map": { |
| "AutoConfig": "configuration_mark_v3.MarkV3Config", |
| "AutoModelForCausalLM": "modeling_mark_v3.MarkV3ForCausalLM" |
| }, |
| "model_type": "deepseek2", |
| "model_name": "Mark-SLM-V3-150M", |
| "model_version": "3.0.0", |
| "vocab_size": 32032, |
| "hidden_size": 576, |
| "num_hidden_layers": 15, |
| "num_attention_heads": 9, |
| "num_key_value_heads": 9, |
| "intermediate_size": 1536, |
| "use_mla": true, |
| "q_lora_rank": 256, |
| "kv_lora_rank": 128, |
| "qk_nope_head_dim": 64, |
| "qk_rope_head_dim": 64, |
| "v_head_dim": 64, |
| "n_routed_experts": 16, |
| "num_experts_per_tok": 2, |
| "n_shared_experts": 1, |
| "moe_intermediate_size": 256, |
| "max_position_embeddings": 8192, |
| "original_max_position_embeddings": 4096, |
| "rope_theta": 10000.0, |
| "rope_scaling_factor": 2.0, |
| "rms_norm_eps": 1e-06, |
| "tie_word_embeddings": true, |
| "torch_dtype": "bfloat16", |
| "quantization": "IQ4_NL", |
| "total_parameters": 146899248, |
| "active_parameters_per_token": 54001968, |
| "weight_state": "TRAINED_CHECKPOINT", |
| "sha256": "4bbfe2dc3fc7a6ac895d56e30352bda0de1322f9826afa28cafe7d24b0e99632", |
| "target_platforms": [ |
| "Ollama", |
| "LM Studio", |
| "llama.cpp", |
| "PyTorch" |
| ] |
| } |
|
|
