Table Question Answering
Transformers
ONNX
Safetensors
English
modernbert
fill-mask
table-grounding
number-verification
clinical-study-report
Instructions to use decosaai/decosa-cell-pointer-modernbert-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use decosaai/decosa-cell-pointer-modernbert-base with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("table-question-answering", model="decosaai/decosa-cell-pointer-modernbert-base")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForMaskedLM tokenizer = AutoTokenizer.from_pretrained("decosaai/decosa-cell-pointer-modernbert-base") model = AutoModelForMaskedLM.from_pretrained("decosaai/decosa-cell-pointer-modernbert-base", device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 3,669 Bytes
2769c0b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 | {
"model": "decosa-cell-pointer-modernbert-base",
"default": "v2-blind (root)",
"twin": "v1-aware (aware/)",
"date": "2026-09-27",
"setup": "CSR number checker end to end, only the pointer changes; BM25 table retrieval on every row; held-out sets never trained on; thresholds and model choice on the model's own dev split",
"baseline": "Qwen3.8-27B as the pointer, values shown",
"sets": {
"synthetic_test": {
"desc": "12 fictional CSRs from the training generator (different seeds), 96 planted errors",
"qwen3.8-27b": {
"caught": 93,
"planted": 96,
"false_flags_per_100_numbers": 0.11,
"right_cell": 0.9898,
"right_cell_planted": 0.8542,
"s_per_100_pages": 79.3
},
"v2-blind": {
"caught": 96,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 1.0,
"right_cell_planted": 1.0,
"s_per_100_pages": 69.1
},
"v1-aware": {
"caught": 96,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 1.0,
"right_cell_planted": 1.0,
"s_per_100_pages": 139
},
"cascade_v2_plus_qwen_below_0.6": {
"caught": 95,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 0.999,
"right_cell_planted": 0.99
}
},
"rewritten_test": {
"desc": "the same reports with paragraphs rewritten by Qwen3.8-27B, 96 planted errors",
"qwen3.8-27b": {
"caught": 93,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 0.9869,
"right_cell_planted": 0.8211
},
"v2-blind": {
"caught": 95,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 1.0,
"right_cell_planted": 1.0,
"s_per_100_pages": 71.5
},
"v1-aware": {
"caught": 95,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 0.999,
"right_cell_planted": 0.989,
"s_per_100_pages": 156
},
"cascade_v2_plus_qwen_below_0.6": {
"caught": 93,
"planted": 96,
"false_flags_per_100_numbers": 0.0,
"right_cell": 0.996,
"right_cell_planted": 0.958
}
},
"clinicaltrials_gov": {
"desc": "30 real phase 3 trials (conditions not in training), posted results as report tables with template narratives, 178 planted errors",
"qwen3.8-27b": {
"caught": 172,
"planted": 178,
"false_flags_per_100_numbers": 2.79,
"right_cell": 0.9717,
"right_cell_planted": 0.8757,
"s_per_100_pages": 307.2
},
"v2-blind": {
"caught": 176,
"planted": 178,
"false_flags_per_100_numbers": 4.08,
"right_cell": 0.9778,
"right_cell_planted": 0.9774,
"s_per_100_pages": 31.3
},
"v1-aware": {
"caught": 176,
"planted": 178,
"false_flags_per_100_numbers": 3.65,
"right_cell": 0.9817,
"right_cell_planted": 0.9774,
"s_per_100_pages": 61.7
},
"cascade_v2_plus_qwen_below_0.6": {
"caught": 176,
"planted": 178,
"false_flags_per_100_numbers": 3.76,
"right_cell": 0.981,
"right_cell_planted": 0.977
}
}
},
"transfer_10k_mdna": {
"desc": "two FY2025 10-Ks not trained on plus a synthetic sample, 69 figures; gold from a value-aware code tie-out",
"right_cell": {
"qwen3.8-27b_values_shown": 0.957,
"qwen3.8-27b_values_hidden": 0.71,
"v1-blind": 0.391,
"v1-aware": 0.435,
"v2-blind": 0.565
},
"note": "v1-aware pointed at the cell holding the written value on a planted swapped-period figure; the blind models did not"
},
"cascade_verdict": "does not help; use the model alone",
"latency": "0.2-0.45 s CPU per checked number, shared server, ONNX Runtime 4 threads",
"source": "docs/evals/cell-pointer.md in the Decosa API repository"
} |