M3TiS-33B: Nanbeige4.3-33B EXL3 3.5 bpw + DFlash2 draft head + TensorFold server (engine aacb6300)
ae9d494 verified | { | |
| "architectures": [ | |
| "DFlash2DraftModel" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": null, | |
| "dflash_config": { | |
| "block_size": 8, | |
| "conv_group_size": 16, | |
| "conv_kernel_size": 2, | |
| "mask_token_id": 1, | |
| "selector_rank": 256, | |
| "selector_top_k": 16, | |
| "target_layer_ids": [ | |
| 4, | |
| 15, | |
| 26, | |
| 38, | |
| 49 | |
| ] | |
| }, | |
| "draft_vocab_size": 166912, | |
| "dtype": "bfloat16", | |
| "eos_token_id": 166101, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 7168, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 11008, | |
| "is_causal": false, | |
| "layer_types": [ | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention" | |
| ], | |
| "max_position_embeddings": 524288, | |
| "max_window_layers": 5, | |
| "model_type": "qwen3", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 5, | |
| "num_key_value_heads": 8, | |
| "num_target_layers": 52, | |
| "pad_token_id": 0, | |
| "rms_norm_eps": 1e-05, | |
| "rope_parameters": { | |
| "rope_theta": 10000000, | |
| "rope_type": "default" | |
| }, | |
| "rope_theta": 10000000, | |
| "sliding_window": 2048, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.12.1", | |
| "use_cache": true, | |
| "use_sliding_window": true, | |
| "vocab_size": 166912 | |
| } | |