{ "vit_hidden_dim": 768, "vit_inter_dim": 3072, "vit_patch_size": 16, "vit_img_size": 224, "vit_n_heads": 12, "vit_dropout": 0.0, "vit_n_blocks": 12, "vit_ln_eps": 1e-06, "vit_cls_flag": false, "vit_model_type": "google/siglip-base-patch16-224", "lm_hidden_dim": 576, "lm_inter_dim": 1536, "lm_rms_eps": 1e-05, "lm_re_base": 100000, "lm_max_position_embeddings": 8192, "lm_base_vocab_size": 49152, "extra_token_amount": 1, "lm_vocab_size": 49153, "lm_n_heads": 9, "lm_n_kv_heads": 3, "lm_dropout": 0.0, "lm_n_blocks": 30, "lm_attn_scaling": 1.0, "lm_eos_token_id": 0, "lm_max_length": 128, "lm_use_tokens": false, "lm_tie_weights": true, "lm_model_type": "HuggingFaceTB/SmolLM2-135M", "lm_tokenizer": "HuggingFaceTB/cosmo2-tokenizer", "mp_pixel_shuffle_factor": 2, "mp_image_token_length": 49, "vlm_extra_tokens": { "image_token": "<|image|>" }, "vlm_load_backbone_weights": true, "vlm_checkpoint_path": "checkpoints", "hf_repo_name": "nanoVLM" }