| { | |
| "architectures": [ | |
| "Grok1ForCausalLM" | |
| ], | |
| "attention_probs_dropout_prob": 0.1, | |
| "attn_factor": 1.0, | |
| "attn_logit_softcapping": 30.0, | |
| "attn_temperature_len": 1024, | |
| "beta_fast": 8, | |
| "beta_slow": 1, | |
| "dtype": "bfloat16", | |
| "embedding_multiplier_scale": 90.50966799187809, | |
| "eos_token_id": 102, | |
| "extrapolation_factor": 1.0, | |
| "final_logit_softcapping": 50, | |
| "global_attn_every_n": 1, | |
| "head_dim": 128, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_size": 8192, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 32768, | |
| "layer_norm_eps": 1e-12, | |
| "max_position_embeddings": 131072, | |
| "model_type": "git", | |
| "moe_intermediate_size": 16384, | |
| "num_attention_heads": 64, | |
| "num_experts_per_tok": 2, | |
| "num_hidden_layers": 64, | |
| "num_image_with_embedding": null, | |
| "num_key_value_heads": 8, | |
| "num_local_experts": 8, | |
| "original_max_position_embeddings": 8192, | |
| "output_multiplier_scale": 0.5, | |
| "pad_token_id": 0, | |
| "position_embedding_type": "absolute", | |
| "residual_moe": true, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 208533496, | |
| "rope_type": "original", | |
| "router_logit_softcapping": 30.0, | |
| "scaling_factor": 16.0, | |
| "sliding_window_size": -1, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.56.1", | |
| "unsloth_fixed": true, | |
| "use_cache": true, | |
| "vision_config": { | |
| "attention_dropout": 0.0, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 768, | |
| "image_size": 224, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "layer_norm_eps": 1e-05, | |
| "model_type": "git_vision_model", | |
| "num_attention_heads": 12, | |
| "num_channels": 3, | |
| "num_hidden_layers": 12, | |
| "patch_size": 16 | |
| }, | |
| "vocab_size": 131072 | |
| } |