| 1 | { |
| 2 | "abhigoyal/vllm-medusa-llama-68m-random": { |
| 3 | "architectures": [ |
| 4 | "MedusaModel" |
| 5 | ], |
| 6 | "model_type": "medusa", |
| 7 | "text_model_type": "medusa", |
| 8 | "hidden_size": 768, |
| 9 | "total_num_hidden_layers": 1, |
| 10 | "total_num_attention_heads": 0, |
| 11 | "head_size": "Error: integer division or modulo by zero", |
| 12 | "vocab_size": 32000, |
| 13 | "total_num_kv_heads": 0, |
| 14 | "num_experts": 0, |
| 15 | "is_deepseek_mla": false, |
| 16 | "is_multimodal_model": false, |
| 17 | "dtype": "torch.float32" |
| 18 | }, |
| 19 | "luccafong/deepseek_mtp_draft_random": { |
| 20 | "architectures": [ |
| 21 | "DeepSeekMTPModel" |
| 22 | ], |
| 23 | "model_type": "deepseek_mtp", |
| 24 | "text_model_type": "deepseek_mtp", |
| 25 | "hidden_size": 2560, |
| 26 | "total_num_hidden_layers": 1, |
| 27 | "total_num_attention_heads": 32, |
| 28 | "head_size": 576, |
| 29 | "vocab_size": 129280, |
| 30 | "total_num_kv_heads": 32, |
| 31 | "num_experts": 72, |
| 32 | "is_deepseek_mla": true, |
| 33 | "is_multimodal_model": false, |
| 34 | "dtype": "torch.bfloat16" |
| 35 | }, |
| 36 | "eagle618/eagle-deepseek-v3-random": { |
| 37 | "architectures": [ |
| 38 | "EagleDeepSeekMTPModel" |
| 39 | ], |
| 40 | "model_type": "eagle", |
| 41 | "text_model_type": "eagle", |
| 42 | "hidden_size": 2560, |
| 43 | "total_num_hidden_layers": 1, |
| 44 | "total_num_attention_heads": 32, |
| 45 | "head_size": 576, |
| 46 | "vocab_size": 129280, |
| 47 | "total_num_kv_heads": 32, |
| 48 | "num_experts": 72, |
| 49 | "is_deepseek_mla": true, |
| 50 | "is_multimodal_model": false, |
| 51 | "dtype": "bfloat16" |
| 52 | }, |
| 53 | "yuhuili/EAGLE-LLaMA3-Instruct-8B": { |
| 54 | "architectures": [ |
| 55 | "EagleLlamaForCausalLM" |
| 56 | ], |
| 57 | "model_type": "eagle", |
| 58 | "text_model_type": "eagle", |
| 59 | "hidden_size": 4096, |
| 60 | "total_num_hidden_layers": 1, |
| 61 | "total_num_attention_heads": 32, |
| 62 | "head_size": 128, |
| 63 | "vocab_size": 128256, |
| 64 | "total_num_kv_heads": 8, |
| 65 | "num_experts": 0, |
| 66 | "is_deepseek_mla": false, |
| 67 | "is_multimodal_model": false, |
| 68 | "dtype": "float16" |
| 69 | }, |
| 70 | "yuhuili/EAGLE3-LLaMA3.1-Instruct-8B": { |
| 71 | "architectures": [ |
| 72 | "Eagle3LlamaForCausalLM" |
| 73 | ], |
| 74 | "model_type": "eagle", |
| 75 | "text_model_type": "eagle", |
| 76 | "hidden_size": 4096, |
| 77 | "total_num_hidden_layers": 1, |
| 78 | "total_num_attention_heads": 32, |
| 79 | "head_size": 128, |
| 80 | "vocab_size": 128256, |
| 81 | "total_num_kv_heads": 8, |
| 82 | "num_experts": 0, |
| 83 | "is_deepseek_mla": false, |
| 84 | "is_multimodal_model": false, |
| 85 | "dtype": "float16" |
| 86 | } |
| 87 | } |