Image-to-Text
Transformers
PyTorch
English
udop
text-generation
chemistry
markush
cxsmiles
molecular-structure
ocr
document-understanding
vision-language-model
patent-analysis
Instructions to use docling-project/MarkushGrapher-2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use docling-project/MarkushGrapher-2 with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "image-to-text" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("image-to-text", model="docling-project/MarkushGrapher-2")# Load model directly from transformers import AutoProcessor, AutoModelForSeq2SeqLM processor = AutoProcessor.from_pretrained("docling-project/MarkushGrapher-2") model = AutoModelForSeq2SeqLM.from_pretrained("docling-project/MarkushGrapher-2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "_name_or_path": "./models/12_15_AM_October_24_2025/checkpoint-100000", | |
| "architecture_variant": "me-lf-stack-1", | |
| "architectures": [ | |
| "MarkushgrapherForConditionalGeneration" | |
| ], | |
| "attention_dropout": null, | |
| "attention_type": "block_sparse", | |
| "attention_window": 512, | |
| "context_embeddings": [], | |
| "context_residual": null, | |
| "context_weight_update_time": 0.5, | |
| "d_ff": 4096, | |
| "d_kv": 64, | |
| "d_model": 1024, | |
| "data_dir": "/mnt/terrantang/data", | |
| "decoder_start_token_id": 0, | |
| "dense_act_fn": "relu", | |
| "disable_sequential_embeddings": false, | |
| "do_lower_case": false, | |
| "dropout_rate": 0.1, | |
| "embedding_1dpos_type": "embed", | |
| "embedding_2dpos_type": "embed", | |
| "embedding_skip_connect": false, | |
| "embedding_use_hw": false, | |
| "embedding_use_mlp": false, | |
| "eos_token_id": 1, | |
| "feed_forward_proj": "relu", | |
| "florence_checkpoint": "Florence-v1.1/florence-v1.1-davit-d3-224x224-64ccc2.pt", | |
| "florence_version": "Florence-v1.1-davit-d3", | |
| "image_size": 512, | |
| "initializer_factor": 1.0, | |
| "is_encoder_decoder": true, | |
| "is_gated_act": false, | |
| "layer_norm_epsilon": 1e-06, | |
| "layernorm_type": "torch_LayerNorm", | |
| "locked_dropout": 0.0, | |
| "mae_checkpoint": "mae_models/mae_pretrain_vit_large_full.pth", | |
| "mae_version": "mae_vit_large_patch16", | |
| "max_2d_position_embeddings": 1024, | |
| "max_bbox_length": 1001, | |
| "max_context_weight": 1.0, | |
| "max_pos_dropout": 0.0, | |
| "max_position_embeddings": 512, | |
| "model_type": "udop", | |
| "n_positions": 512, | |
| "num_channels": 3, | |
| "num_decoder_layers": 24, | |
| "num_doc_classes": 40, | |
| "num_heads": 16, | |
| "num_layers": 24, | |
| "output_attentions": true, | |
| "output_past": true, | |
| "pad_token_id": 0, | |
| "page_embeddings_args": {}, | |
| "page_embeddings_type": "none", | |
| "patch_size": 16, | |
| "pos_dropout_update_time": 0.5, | |
| "positional_dropout_type": "random", | |
| "relative_attention_max_distance": 128, | |
| "relative_attention_num_buckets": 32, | |
| "relative_bias_args": [ | |
| { | |
| "type": "1d" | |
| }, | |
| { | |
| "type": "horizontal" | |
| }, | |
| { | |
| "type": "vertical" | |
| } | |
| ], | |
| "schema_version": null, | |
| "syncbn_type": "torch_SyncBatchNorm", | |
| "task_specific_params": { | |
| "summarization": { | |
| "early_stopping": true, | |
| "length_penalty": 2.0, | |
| "max_length": 200, | |
| "min_length": 30, | |
| "no_repeat_ngram_size": 3, | |
| "num_beams": 4, | |
| "prefix": "summarize: " | |
| }, | |
| "translation_en_to_de": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to German: " | |
| }, | |
| "translation_en_to_fr": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to French: " | |
| }, | |
| "translation_en_to_ro": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to Romanian: " | |
| } | |
| }, | |
| "torch_dtype": "float32", | |
| "training_tasks": [ | |
| "lm" | |
| ], | |
| "transformers_version": "4.34.0.dev0", | |
| "truncate_decoder_after_layer": null, | |
| "truncate_encoder_after_layer": null, | |
| "use_cache": true, | |
| "vision_augmentation": [], | |
| "vocab_size": 33201, | |
| "word_dropout": 0.0 | |
| } | |