CoolFace
Modelpublic

Pinaster/Kimi-K2.5-4layer

sourceHugging Faceotherupdated 6mo agoView on Hugging Face
0likes30downloads
config.json192 linesDownload Raw Back to root
1{2  "architectures": [3    "KimiK25ForConditionalGeneration"4  ],5  "auto_map": {6    "AutoConfig": "configuration_kimi_k25.KimiK25Config",7    "AutoModel": "modeling_kimi_k25.KimiK25ForConditionalGeneration",8    "AutoModelForCausalLM": "modeling_kimi_k25.KimiK25ForConditionalGeneration"9  },10  "bos_token_id": 163584,11  "dtype": "bfloat16",12  "eos_token_id": 163585,13  "ignore_index": -100,14  "media_placeholder_token_id": 163605,15  "model_type": "kimi_k25",16  "pad_token_id": 163839,17  "text_config": {18    "_name_or_path": "",19    "add_cross_attention": false,20    "architectures": [21      "DeepseekV3ForCausalLM"22    ],23    "attention_bias": false,24    "attention_dropout": 0.0,25    "auto_map": {26      "AutoConfig": "configuration_deepseek.DeepseekV3Config",27      "AutoModel": "modeling_deepseek.DeepseekV3Model",28      "AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM"29    },30    "aux_loss_alpha": 0.001,31    "bad_words_ids": null,32    "begin_suppress_tokens": null,33    "bos_token_id": 163584,34    "chunk_size_feed_forward": 0,35    "cross_attention_hidden_size": null,36    "decoder_start_token_id": null,37    "diversity_penalty": 0.0,38    "do_sample": false,39    "dtype": "bfloat16",40    "early_stopping": false,41    "encoder_no_repeat_ngram_size": 0,42    "eos_token_id": 163585,43    "ep_size": 1,44    "exponential_decay_length_penalty": null,45    "finetuning_task": null,46    "first_k_dense_replace": 1,47    "forced_bos_token_id": null,48    "forced_eos_token_id": null,49    "hidden_act": "silu",50    "hidden_size": 7168,51    "id2label": {52      "0": "LABEL_0",53      "1": "LABEL_1"54    },55    "initializer_range": 0.02,56    "intermediate_size": 18432,57    "is_decoder": false,58    "is_encoder_decoder": false,59    "kv_lora_rank": 512,60    "label2id": {61      "LABEL_0": 0,62      "LABEL_1": 163    },64    "length_penalty": 1.0,65    "max_length": 20,66    "max_position_embeddings": 262144,67    "min_length": 0,68    "model_type": "kimi_k2",69    "moe_intermediate_size": 2048,70    "moe_layer_freq": 1,71    "n_group": 1,72    "n_routed_experts": 384,73    "n_shared_experts": 1,74    "no_repeat_ngram_size": 0,75    "norm_topk_prob": true,76    "num_attention_heads": 64,77    "num_beam_groups": 1,78    "num_beams": 1,79    "num_experts_per_tok": 8,80    "num_hidden_layers": 4,81    "num_key_value_heads": 64,82    "num_nextn_predict_layers": 0,83    "num_return_sequences": 1,84    "output_attentions": false,85    "output_hidden_states": false,86    "output_scores": false,87    "pad_token_id": 163839,88    "prefix": null,89    "pretraining_tp": 1,90    "problem_type": null,91    "pruned_heads": {},92    "q_lora_rank": 1536,93    "qk_nope_head_dim": 128,94    "qk_rope_head_dim": 64,95    "quantization_config": {96      "config_groups": {97        "group_0": {98          "input_activations": null,99          "output_activations": null,100          "targets": [101            "Linear"102          ],103          "weights": {104            "actorder": null,105            "block_structure": null,106            "dynamic": false,107            "group_size": 32,108            "num_bits": 4,109            "observer": "minmax",110            "observer_kwargs": {},111            "strategy": "group",112            "symmetric": true,113            "type": "int"114          }115        }116      },117      "format": "pack-quantized",118      "ignore": [119        "lm_head",120        "re:.*self_attn.*",121        "re:.*shared_experts.*",122        "re:.*mlp\\.(gate|up|gate_up|down)_proj.*"123      ],124      "kv_cache_scheme": null,125      "quant_method": "compressed-tensors",126      "quantization_status": "compressed"127    },128    "remove_invalid_values": false,129    "repetition_penalty": 1.0,130    "return_dict": true,131    "return_dict_in_generate": false,132    "rms_norm_eps": 1e-05,133    "rope_scaling": {134      "beta_fast": 32.0,135      "beta_slow": 1.0,136      "factor": 64.0,137      "mscale": 1.0,138      "mscale_all_dim": 1.0,139      "original_max_position_embeddings": 4096,140      "type": "yarn"141    },142    "rope_theta": 50000.0,143    "routed_scaling_factor": 2.827,144    "scoring_func": "sigmoid",145    "sep_token_id": null,146    "seq_aux": true,147    "suppress_tokens": null,148    "task_specific_params": null,149    "temperature": 1.0,150    "tf_legacy_loss": false,151    "tie_encoder_decoder": false,152    "tie_word_embeddings": false,153    "tokenizer_class": null,154    "top_k": 50,155    "top_p": 1.0,156    "topk_group": 1,157    "topk_method": "noaux_tc",158    "torchscript": false,159    "transformers_version": "4.56.2",160    "typical_p": 1.0,161    "use_bfloat16": false,162    "use_cache": true,163    "v_head_dim": 128,164    "vocab_size": 163840165  },166  "tie_word_embeddings": false,167  "use_unified_vision_chunk": true,168  "video_placeholder": "<|kimi_k25_video_placeholder|>",169  "vision_config": {170    "_attn_implementation": "flash_attention_2",171    "init_pos_emb_height": 64,172    "init_pos_emb_time": 4,173    "init_pos_emb_width": 64,174    "merge_kernel_size": [175      2,176      2177    ],178    "merge_type": "sd2_tpool",179    "mm_hidden_size": 1152,180    "mm_projector_type": "patchmerger",181    "patch_size": 14,182    "pos_emb_type": "divided_fixed",183    "projector_hidden_act": "gelu",184    "projector_ln_eps": 1e-05,185    "text_hidden_size": 7168,186    "video_attn_type": "spatial_temporal",187    "vt_hidden_size": 1152,188    "vt_intermediate_size": 4304,189    "vt_num_attention_heads": 16,190    "vt_num_hidden_layers": 27191  }192}