Zero-Shot Classification
Transformers
Safetensors
English
qwen3_5_text
text-generation
scientific-papers
taxonomy
classification
science-of-science
probabilities
qwen3_5
Instructions to use maximerivest/paper-classifier-2B-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use maximerivest/paper-classifier-2B-v2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("zero-shot-classification", model="maximerivest/paper-classifier-2B-v2")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("maximerivest/paper-classifier-2B-v2") model = AutoModelForCausalLM.from_pretrained("maximerivest/paper-classifier-2B-v2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download benchmarks.json from maximerivest/paper-classifier-2B-v2: direct link, hf CLI and curl.
- Browser
- Download file 12.3 kB
-
https://huggingface.co/maximerivest/paper-classifier-2B-v2/resolve/main/benchmarks.json
- Command line
-
hf download hf://maximerivest/paper-classifier-2B-v2/benchmarks.json
-
curl -L -o benchmarks.json https://huggingface.co/maximerivest/paper-classifier-2B-v2/resolve/main/benchmarks.json
12.3 kB
| { | |
| "schema_version": 1, | |
| "model": "maximerivest/paper-classifier-2B-v2", | |
| "expert_200": { | |
| "expert_flat": { | |
| "disciplinary": [ | |
| 57.0, | |
| 47.2, | |
| 66.3 | |
| ], | |
| "multidisciplinary": [ | |
| 66.0, | |
| 56.3, | |
| 74.5 | |
| ], | |
| "top3": 86.0, | |
| "ece_vs_experts": 0.046, | |
| "mean_confidence": 0.649 | |
| }, | |
| "expert_flat_compact": { | |
| "disciplinary": [ | |
| 61.0, | |
| 51.2, | |
| 70.0 | |
| ], | |
| "multidisciplinary": [ | |
| 65.0, | |
| 55.3, | |
| 73.6 | |
| ], | |
| "top3": 85.5 | |
| }, | |
| "expert_shuffled": { | |
| "disciplinary": [ | |
| 57.0, | |
| 47.2, | |
| 66.3 | |
| ], | |
| "multidisciplinary": [ | |
| 64.0, | |
| 54.2, | |
| 72.7 | |
| ], | |
| "same_pick_as_unshuffled": 76.5 | |
| }, | |
| "expert_field": { | |
| "field_agreement": [ | |
| 51.5, | |
| 44.6, | |
| 58.3 | |
| ] | |
| }, | |
| "expert_tree": { | |
| "disciplinary": [ | |
| 47.0, | |
| 37.5, | |
| 56.7 | |
| ], | |
| "multidisciplinary": [ | |
| 33.0, | |
| 24.6, | |
| 42.7 | |
| ] | |
| }, | |
| "yesno_heldout_topics": { | |
| "macro_average_precision": 0.8196, | |
| "macro_auc": 0.9939, | |
| "topics": 8 | |
| }, | |
| "unseen_invented": { | |
| "reference_a": { | |
| "agree_top": 81.6, | |
| "log_loss": 0.614, | |
| "ece": 0.0206, | |
| "n": 800 | |
| }, | |
| "reference_b": { | |
| "agree_top": 81.1, | |
| "log_loss": 0.702, | |
| "ece": 0.0313, | |
| "n": 760 | |
| }, | |
| "reference_agreement": 89.9 | |
| }, | |
| "unseen_jel": { | |
| "reference_a": { | |
| "agree_top": 82.8, | |
| "log_loss": 0.558, | |
| "ece": 0.0233, | |
| "n": 600 | |
| }, | |
| "reference_b": { | |
| "agree_top": 83.7, | |
| "log_loss": 0.809, | |
| "ece": 0.0217, | |
| "n": 600 | |
| }, | |
| "reference_agreement": 91.8 | |
| }, | |
| "unseen_papers": { | |
| "reference_a": { | |
| "agree_top": 87.4, | |
| "log_loss": 0.476, | |
| "ece": 0.0337, | |
| "n": 1200 | |
| }, | |
| "reference_b": { | |
| "agree_top": 87.7, | |
| "log_loss": 0.589, | |
| "ece": 0.0371, | |
| "n": 1180 | |
| }, | |
| "reference_agreement": 91.5 | |
| } | |
| }, | |
| "arxiv_and_mesh": { | |
| "arxiv": { | |
| "flat": { | |
| "primary": [ | |
| 57.3, | |
| 55.6, | |
| 59.1 | |
| ], | |
| "any_category": [ | |
| 70.4, | |
| 68.8, | |
| 72.0 | |
| ], | |
| "top3": [ | |
| 82.5, | |
| 81.1, | |
| 83.8 | |
| ], | |
| "group_from_pick": [ | |
| 85.4, | |
| 84.1, | |
| 86.6 | |
| ], | |
| "ece": 0.0793, | |
| "mean_confidence": 0.652, | |
| "n": 2980 | |
| }, | |
| "about": { | |
| "primary": [ | |
| 55.8, | |
| 54.0, | |
| 57.5 | |
| ], | |
| "any_category": [ | |
| 69.3, | |
| 67.6, | |
| 70.9 | |
| ], | |
| "top3": [ | |
| 82.2, | |
| 80.8, | |
| 83.5 | |
| ], | |
| "group_from_pick": [ | |
| 84.6, | |
| 83.3, | |
| 85.9 | |
| ], | |
| "ece": 0.0935, | |
| "mean_confidence": 0.651, | |
| "n": 2980 | |
| }, | |
| "title": { | |
| "primary": [ | |
| 46.9, | |
| 45.1, | |
| 48.7 | |
| ], | |
| "any_category": [ | |
| 59.2, | |
| 57.5, | |
| 61.0 | |
| ], | |
| "top3": [ | |
| 69.9, | |
| 68.2, | |
| 71.5 | |
| ], | |
| "group_from_pick": [ | |
| 81.2, | |
| 79.8, | |
| 82.6 | |
| ], | |
| "ece": 0.1444, | |
| "mean_confidence": 0.613, | |
| "n": 2980 | |
| }, | |
| "shuffled": { | |
| "primary": [ | |
| 57.3, | |
| 55.6, | |
| 59.1 | |
| ], | |
| "any_category": [ | |
| 70.0, | |
| 68.3, | |
| 71.6 | |
| ], | |
| "top3": [ | |
| 82.3, | |
| 80.9, | |
| 83.6 | |
| ], | |
| "group_from_pick": [ | |
| 86.4, | |
| 85.1, | |
| 87.6 | |
| ], | |
| "ece": 0.0836, | |
| "mean_confidence": 0.656, | |
| "n": 2980 | |
| }, | |
| "group_direct": { | |
| "accuracy": [ | |
| 80.6, | |
| 79.1, | |
| 82.0 | |
| ], | |
| "ece": 0.1108 | |
| }, | |
| "order": { | |
| "same_pick_shuffled_vs_flat": [ | |
| 81.8, | |
| 80.4, | |
| 83.1 | |
| ], | |
| "conf_0-0.4": { | |
| "n": 429, | |
| "same_pick": 0.522, | |
| "accuracy": 0.315 | |
| }, | |
| "conf_0.4-0.6": { | |
| "n": 863, | |
| "same_pick": 0.694, | |
| "accuracy": 0.417 | |
| }, | |
| "conf_0.6-0.8": { | |
| "n": 782, | |
| "same_pick": 0.914, | |
| "accuracy": 0.587 | |
| }, | |
| "conf_0.8-1.01": { | |
| "n": 906, | |
| "same_pick": 0.992, | |
| "accuracy": 0.833 | |
| } | |
| }, | |
| "flat_by_group": { | |
| "Computer Science": [ | |
| 51.1, | |
| 47.5, | |
| 54.6 | |
| ], | |
| "Economics": [ | |
| 56.7, | |
| 44.1, | |
| 68.4 | |
| ], | |
| "Electrical Engineering and Systems Science": [ | |
| 61.3, | |
| 50.3, | |
| 71.2 | |
| ], | |
| "Mathematics": [ | |
| 66.5, | |
| 62.6, | |
| 70.2 | |
| ], | |
| "Physics": [ | |
| 63.2, | |
| 60.2, | |
| 66.1 | |
| ], | |
| "Quantitative Biology": [ | |
| 34.0, | |
| 27.8, | |
| 40.8 | |
| ], | |
| "Quantitative Finance": [ | |
| 52.5, | |
| 44.8, | |
| 60.1 | |
| ], | |
| "Statistics": [ | |
| 42.0, | |
| 32.8, | |
| 51.8 | |
| ] | |
| }, | |
| "flat_by_year": { | |
| "2008-2011": [ | |
| 54.7, | |
| 49.5, | |
| 59.8 | |
| ], | |
| "2012-2015": [ | |
| 57.3, | |
| 53.7, | |
| 60.8 | |
| ], | |
| "2016-2019": [ | |
| 57.7, | |
| 54.2, | |
| 61.2 | |
| ], | |
| "2020-2023": [ | |
| 58.3, | |
| 54.8, | |
| 61.7 | |
| ], | |
| "2024-2027": [ | |
| 57.2, | |
| 52.1, | |
| 62.1 | |
| ] | |
| }, | |
| "flat_by_in_journal": { | |
| "False": [ | |
| 55.9, | |
| 53.8, | |
| 57.9 | |
| ], | |
| "True": [ | |
| 61.5, | |
| 58.0, | |
| 64.8 | |
| ] | |
| }, | |
| "flat_by_cited_10plus": { | |
| "False": [ | |
| 57.3, | |
| 55.5, | |
| 59.1 | |
| ], | |
| "True": [ | |
| 59.0, | |
| 49.5, | |
| 68.0 | |
| ] | |
| }, | |
| "flat_by_confidence": { | |
| "0.0": [ | |
| 28.6, | |
| 11.7, | |
| 54.6 | |
| ], | |
| "0.2": [ | |
| 31.6, | |
| 27.3, | |
| 36.2 | |
| ], | |
| "0.4": [ | |
| 41.7, | |
| 38.5, | |
| 45.0 | |
| ], | |
| "0.6": [ | |
| 58.7, | |
| 55.2, | |
| 62.1 | |
| ], | |
| "0.8": [ | |
| 83.3, | |
| 80.8, | |
| 85.6 | |
| ] | |
| }, | |
| "flat_worst_categories": { | |
| "cs.OH": [ | |
| 0.0, | |
| -0.0, | |
| 16.1 | |
| ], | |
| "q-bio.SC": [ | |
| 0.0, | |
| -0.0, | |
| 16.1 | |
| ], | |
| "cs.GL": [ | |
| 5.0, | |
| 0.9, | |
| 23.6 | |
| ], | |
| "q-bio.BM": [ | |
| 5.0, | |
| 0.9, | |
| 23.6 | |
| ], | |
| "q-bio.OT": [ | |
| 5.0, | |
| 0.9, | |
| 23.6 | |
| ], | |
| "cs.DM": [ | |
| 10.0, | |
| 2.8, | |
| 30.1 | |
| ], | |
| "cs.ET": [ | |
| 10.0, | |
| 2.8, | |
| 30.1 | |
| ], | |
| "cs.PF": [ | |
| 10.0, | |
| 2.8, | |
| 30.1 | |
| ], | |
| "physics.data-an": [ | |
| 10.0, | |
| 2.8, | |
| 30.1 | |
| ], | |
| "cond-mat.other": [ | |
| 15.0, | |
| 5.2, | |
| 36.0 | |
| ], | |
| "cs.CE": [ | |
| 15.0, | |
| 5.2, | |
| 36.0 | |
| ], | |
| "q-bio.QM": [ | |
| 15.0, | |
| 5.2, | |
| 36.0 | |
| ], | |
| "cs.MM": [ | |
| 20.0, | |
| 8.1, | |
| 41.6 | |
| ], | |
| "math-ph": [ | |
| 25.0, | |
| 11.2, | |
| 46.9 | |
| ], | |
| "physics.pop-ph": [ | |
| 25.0, | |
| 11.2, | |
| 46.9 | |
| ] | |
| }, | |
| "flat_confusions": [ | |
| "q-bio.SC -> physics.bio-ph: 14", | |
| "stat.ML -> cs.LG: 10", | |
| "q-bio.BM -> physics.bio-ph: 9", | |
| "cs.DM -> math.CO: 7", | |
| "cs.OS -> cs.DC: 7", | |
| "q-bio.CB -> q-bio.QM: 7", | |
| "cs.PF -> cs.DC: 7", | |
| "q-bio.TO -> physics.bio-ph: 6", | |
| "physics.atm-clus -> physics.chem-ph: 6", | |
| "cs.PL -> cs.LO: 6", | |
| "cs.SI -> physics.soc-ph: 6", | |
| "q-bio.BM -> q-bio.QM: 6", | |
| "cs.PF -> cs.NI: 5", | |
| "math.GN -> math.LO: 5", | |
| "econ.EM -> stat.AP: 5", | |
| "math.GM -> math.NT: 5", | |
| "cs.NI -> cs.DC: 5", | |
| "econ.GN -> econ.TH: 5", | |
| "physics.space-ph -> astro-ph.EP: 4", | |
| "cs.GR -> cs.HC: 4" | |
| ], | |
| "flat_cross_group_errors": { | |
| "same_group": 848, | |
| "other_group": 423 | |
| } | |
| }, | |
| "mesh": { | |
| "yesno_all": { | |
| "auc": 0.8372, | |
| "average_precision": 0.5485, | |
| "accuracy_at_0.5": 0.7715, | |
| "recall_major_at_0.5": 0.68, | |
| "ece": 0.0266, | |
| "n": 9826, | |
| "positives": 2500 | |
| }, | |
| "yesno_major_vs_minor": { | |
| "auc": 0.6069, | |
| "average_precision": 0.575, | |
| "false_positive_rate_at_0.5": 0.52, | |
| "n_negatives": 2500 | |
| }, | |
| "yesno_major_vs_sibling": { | |
| "auc": 0.9324, | |
| "average_precision": 0.9357, | |
| "false_positive_rate_at_0.5": 0.0505, | |
| "n_negatives": 2375 | |
| }, | |
| "yesno_major_vs_branch": { | |
| "auc": 0.9799, | |
| "average_precision": 0.9769, | |
| "false_positive_rate_at_0.5": 0.0102, | |
| "n_negatives": 2451 | |
| }, | |
| "yesno_auc_by_category": { | |
| "A": 0.864, | |
| "B": 0.861, | |
| "C": 0.883, | |
| "D": 0.836, | |
| "E": 0.841, | |
| "F": 0.828, | |
| "G": 0.856, | |
| "H": 0.886, | |
| "I": 0.823, | |
| "J": 0.8, | |
| "K": 0.832, | |
| "L": 0.791, | |
| "M": 0.807, | |
| "N": 0.82 | |
| }, | |
| "yesno_recall_by_n_majors": { | |
| "1": [ | |
| 69.2, | |
| 66.4, | |
| 71.8 | |
| ], | |
| "2": [ | |
| 69.4, | |
| 66.1, | |
| 72.4 | |
| ], | |
| "3": [ | |
| 65.9, | |
| 60.8, | |
| 70.7 | |
| ], | |
| "4": [ | |
| 58.8, | |
| 51.7, | |
| 65.6 | |
| ] | |
| }, | |
| "pick": { | |
| "accuracy_any_major": [ | |
| 45.0, | |
| 42.9, | |
| 47.2 | |
| ], | |
| "by_n_options": { | |
| "10": [ | |
| 28.7, | |
| 21.7, | |
| 36.8 | |
| ], | |
| "11": [ | |
| 33.7, | |
| 25.0, | |
| 43.7 | |
| ], | |
| "12": [ | |
| 40.8, | |
| 30.4, | |
| 52.0 | |
| ], | |
| "4": [ | |
| 58.5, | |
| 53.1, | |
| 63.8 | |
| ], | |
| "5": [ | |
| 49.1, | |
| 44.0, | |
| 54.1 | |
| ], | |
| "6": [ | |
| 48.1, | |
| 42.9, | |
| 53.4 | |
| ], | |
| "7": [ | |
| 38.7, | |
| 33.3, | |
| 44.4 | |
| ], | |
| "8": [ | |
| 39.3, | |
| 33.1, | |
| 45.8 | |
| ], | |
| "9": [ | |
| 43.6, | |
| 36.6, | |
| 50.9 | |
| ] | |
| }, | |
| "chance": 0.307, | |
| "ece": 0.1792 | |
| } | |
| } | |
| }, | |
| "original_topic_judging": { | |
| "frag": { | |
| "population": { | |
| "auc": 0.9702, | |
| "agree": 0.9602, | |
| "agree_sample": 0.9054, | |
| "precision": 0.708, | |
| "recall": 0.808, | |
| "f1": 0.755, | |
| "average_precision": 0.8342379000300797 | |
| }, | |
| "cited50": { | |
| "auc": 0.9909, | |
| "agree": 0.9798, | |
| "agree_sample": 0.8917, | |
| "precision": 0.732, | |
| "recall": 0.906, | |
| "f1": 0.81 | |
| }, | |
| "meta_refs": { | |
| "found": 0.912, | |
| "n": 640, | |
| "found_title_only": 0.915 | |
| } | |
| }, | |
| "llm": { | |
| "population": { | |
| "auc": 0.9882, | |
| "agree": 0.9548, | |
| "agree_sample": 0.7907, | |
| "precision": 0.51, | |
| "recall": 0.94, | |
| "f1": 0.662, | |
| "average_precision": 0.8262713035326974 | |
| }, | |
| "landmarks": { | |
| "found": 0.932, | |
| "n": 44 | |
| } | |
| } | |
| }, | |
| "notes": [ | |
| "Reference-labelled topic tests are not human ground truth.", | |
| "Original two topics are reused development diagnostics.", | |
| "Paired comparisons in arxiv_and_mesh refer to a private previous list-first checkpoint, not the earlier public paper-first release.", | |
| "Benchmarks use native model prompts and restricted answer-code softmax; no text generation." | |
| ] | |
| } | |