CoolFace
Modelpublic

hf-tiny-v2/tiny-random-Kimi_K25Model

sourceHugging Faceupdated 1mo agoView on Hugging Face
0likes11downloads
config.json81 linesDownload Raw Back to root
1{2  "architectures": [3    "Kimi_K25Model"4  ],5  "dtype": "float32",6  "image_token_id": 3,7  "model_type": "kimi_k25",8  "pad_token_id": 163606,9  "projection_hidden_size": 32,10  "projection_layer_norm_eps": 1e-05,11  "text_config": {12    "attention_bias": false,13    "attention_dropout": 0.0,14    "bos_token_id": 163584,15    "eos_token_id": 163585,16    "first_k_dense_replace": 1,17    "head_dim": 16,18    "hidden_act": "silu",19    "hidden_size": 32,20    "initializer_range": 0.02,21    "intermediate_size": 32,22    "kv_lora_rank": 16,23    "max_position_embeddings": 512,24    "model_type": "deepseek_v3",25    "moe_intermediate_size": 16,26    "n_group": 2,27    "n_routed_experts": 8,28    "n_shared_experts": 1,29    "norm_topk_prob": true,30    "num_attention_heads": 2,31    "num_experts_per_tok": 8,32    "num_hidden_layers": 2,33    "num_key_value_heads": 2,34    "num_nextn_predict_layers": 1,35    "pad_token_id": 163606,36    "pretraining_tp": 1,37    "q_lora_rank": 32,38    "qk_head_dim": 48,39    "qk_nope_head_dim": 32,40    "qk_rope_head_dim": 16,41    "rms_norm_eps": 1e-06,42    "rope_interleave": true,43    "rope_parameters": {44      "rope_theta": 10000,45      "rope_type": "default"46    },47    "routed_scaling_factor": 2.5,48    "tie_word_embeddings": false,49    "topk_group": 1,50    "use_cache": true,51    "v_head_dim": 32,52    "vocab_size": 16360753  },54  "tie_word_embeddings": false,55  "transformers_version": "5.16.0.dev0",56  "video_token_id": 4,57  "vision_config": {58    "hidden_act": "silu",59    "hidden_size": 32,60    "intermediate_size": 32,61    "max_position_embeddings": 512,62    "merge_kernel_size": [63      1,64      165    ],66    "model_type": "kimi_k25_vision",67    "num_attention_heads": 2,68    "num_hidden_layers": 2,69    "patch_size": 8,70    "pos_emb_height": 4,71    "pos_emb_time": 1,72    "pos_emb_width": 4,73    "rope_parameters": {74      "rope_theta": 10000,75      "rope_type": "default"76    }77  },78  "vision_end_token_id": 163604,79  "vision_start_token_id": 16360280}81